How QA Teams Are Saving Hours Every Sprint Using Claude with MCP?
QA teams are cutting hours of manual work every sprint by connecting Claude to Pcloudy through MCP (Model Context Protocol). This video walks through exactly how that integration works —…
About this video
QA teams are cutting hours of manual work every sprint by connecting Claude to Pcloudy through MCP (Model Context Protocol). This video walks through exactly how that integration works — and the time savings it unlocks for real testing workflows.
✅ Step-by-step: connecting Claude to Pcloudy via MCP
✅ Real use cases where teams save hours per sprint
✅ AI-assisted testing without leaving your existing workflow
✅ Practical setup — not just theory
See how AI is actually changing day-to-day QA work
Video Transcript
Full transcript of this video — How QA Teams Are Saving Hours Every Sprint Using Claude with MCP?. Jump to any moment or read straight through.
Dinakar
Hello, everyone, and welcome to pCloudy’s webinar, “How QA Teams Are Saving Hours Every Sprint Using Claude with Qpilot MCP.”
Dinakar
My name is Dinakar, and I’ll be your host today.
Dinakar
Before we begin, I’d like to go over a few housekeeping points.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
If you have any questions or need clarification, please use the Q&A option at the bottom of your screen.
Dinakar
We’ll answer your questions during the Q&A session toward the end of the webinar.
Dinakar
Please also note that we’re recording this webinar.
Dinakar
We’ll share the recording afterward so you can watch it again or share it with your colleagues.
Dinakar
With that said, let me introduce today’s speaker, Jessica Selvin.
Dinakar
Jessica is Lead, Solution Engineering at pCloudy.
Dinakar
She has more than five years of experience in the IT industry.
Dinakar
Before joining pCloudy, she worked at LTIMindtree as a presales consultant, designing automation solutions for various industries.
Dinakar
She also has expertise in data analytics, machine learning, and artificial intelligence tools.
Dinakar
Without further ado, I’ll hand it over to Jessica.
Dinakar
Over to you, Jessica.
Jessica
Thank you so much for that introduction, Dinakar.
Jessica
I hope my screen is visible.
Jessica
As Dinakar mentioned, today’s topic is how Claude and the Qpilot MCP connection can help you generate tests faster, view all your test results in one place, and save hours otherwise spent switching between tools.
Jessica
Let’s begin by looking at what AI has already solved and why some testing activities still take so much time.
Jessica
Test generation, for example, has largely been solved by recent advances in AI.
Jessica
Whether you use an AI agent such as Qpilot or GitHub Copilot to write Playwright scripts, you can describe a flow and generate scripts quickly.
Jessica
That’s genuinely impressive, and the capability has become widely available very quickly.
Jessica
However, most QA teams still use several different tools.
Jessica
A QA engineer might use one tool to write a test, another to share the result, an AI assistant, an editor or IDE, a device farm, and separate dashboards and reports.
Jessica
They may then log bugs or tasks in Slack or Jira.
Jessica
This creates a significant amount of context switching.
Jessica
If API testing is involved, the engineer may also use a tool such as Postman.
Jessica
Debugging a single failure can require at least three or four manual handoffs.
Jessica
Each action and result must also be recorded separately in the relevant tool.
Jessica
This is where teams lose hours during every sprint or release cycle.
Jessica
As people switch between tools, important details are often lost, so their AI tools never receive the full context.
Jessica
Teams also spend time manually reviewing logs, dashboards, and reports.
Jessica
Meanwhile, much of their innovative work is happening inside AI tools such as Cursor or Claude Code.
Jessica
They can’t easily bring the context from other tools back into the environment they primarily use to develop scripts and generate new builds.
Jessica
That is the central problem the Qpilot and Claude MCP connection is designed to solve.
Jessica
Let me summarize the key agenda for today’s session.
Jessica
First, you’ll learn how Claude connects with Qpilot through MCP.
Jessica
You’ll see how to access pCloudy’s real devices directly from tools such as Cursor or Claude.
Jessica
Second, we’ll walk through a live demonstration.
Jessica
Third, I’ll cover our AI-powered test failure analysis reports and show how they significantly reduce debugging time.
Jessica
You’ll see how the system analyzes logs and provides detailed findings quickly.
Jessica
I’ll also explain how to execute test cases and create a reusable MCP setup for your team.
Jessica
As mentioned earlier, we’ll answer questions at the end of the demonstration.
Jessica
Let’s begin with the first section: where QA teams are still losing time and context between tools.
Jessica
In a typical workflow, you often begin with an AI assistant such as Cursor, GitHub Copilot, or Claude Code.
Jessica
You describe the flow you want to test, refine the output through a few exchanges, and eventually receive a script that is ready to run.
Jessica
Next, you copy the script from the chat window and paste it into your codebase.
Jessica
You then fix imports, apply page-object conventions, and replace locators where necessary.
Jessica
This can create further back-and-forth work in your IDE or test editor.
Jessica
When it’s time to use a device farm, you must book a device, add the required capabilities, connect to the device, and run your test cases.
Jessica
You may switch between your IDE or CI/CD tool and a device farm such as pCloudy to track execution.
Jessica
You may also open a separate dashboarding solution to review the results.
Jessica
Finally, you update documentation or log bugs in tools such as Slack or Jira.
Jessica
In this workflow, you’ve switched between five tools and created five opportunities to lose context.
Jessica
You may lose critical details or fixes made in the test editor and fail to bring that context back into your AI assistant.
Jessica
This limits your ability to improve the accuracy of the scripts it generates.
Jessica
That lost context eventually translates into a significant loss of time.
Jessica
Context switching is only one part of the problem.
Jessica
Teams also spend time booking devices, uploading builds, and tracking execution status.
Jessica
Each time, you may need to open a dashboard or active session to check whether execution has finished and whether the report has been generated.
Jessica
These may seem like simple tasks, but they accumulate and consume a meaningful amount of time.
Jessica
Another time-consuming task is reviewing raw logs.
Jessica
When a test case fails, you may need to examine extensive logs to understand what went wrong.
Jessica
With pCloudy’s AI failure analysis agent, you can immediately see the relevant log lines and identify where the test failed.
Jessica
Reporting can create another delay, especially when test results are stored in a separate tool.
Jessica
You still need to retrieve the report, debug the issue, and send the findings back to Jira or Slack.
Jessica
These are the main touchpoints where teams lose both time and context.
Jessica
Now, let me introduce the Claude and Qpilot MCP connection.
Jessica
Claude is the focus of today’s demonstration, but pCloudy also integrates with tools such as Cursor.
Jessica
MCP stands for Model Context Protocol.
Jessica
It attracted considerable attention a few months ago and is an open standard introduced by Anthropic.
Jessica
Put simply, MCP allows an AI assistant to communicate directly with an external system instead of requiring you to act as the intermediary.
Jessica
It enables the assistant to retrieve data from your tools and perform actions on your platform.
Jessica
That is the core idea behind an MCP connection.
Jessica
It provides a common protocol for interacting with different tools and simplifies those connections.
Jessica
You can think of it as an API-like connection between Claude and pCloudy.
Jessica
Instead of reading information in one system, retyping it into another, and carrying the context in your head, MCP connects the systems directly.
Jessica
It removes you from the relay loop and lets your AI assistant communicate directly with the platform.
Jessica
If you haven’t heard of Qpilot, it is pCloudy’s AI-based test creation solution.
Jessica
You can explore our other Qpilot webinars, watch demonstrations on our website, or contact us to learn more.
Jessica
Qpilot is an AI-powered test creation platform built on top of pCloudy’s real-device infrastructure.
Jessica
pCloudy provides real Android and iOS devices, as well as a browser cloud, for manual and automated mobile and web testing.
Jessica
You can run tests using Appium, Selenium, Playwright, and other popular frameworks.
Jessica
Alternatively, you can use Qpilot, our AI layer, to convert plain-English descriptions into automated test steps.
Jessica
Real-device infrastructure is the foundation of the pCloudy platform.
Jessica
We host physical devices, not simulators or emulators, in our device cloud.
Jessica
Through Claude, you can access both Qpilot’s AI capabilities and pCloudy’s real-device infrastructure.
Jessica
Let’s look at some of the practical tasks this connection can support.
Jessica
First, you can generate tests directly from Claude instead of opening Qpilot separately.
Jessica
Once the pCloudy MCP connection is enabled, you can describe the test you need, ask Claude to create it, or retrieve an existing test case.
Jessica
You can also retrieve results and reports, and monitor live executions.
Jessica
This applies to both automated and manual sessions.
Jessica
Because Claude can access those reports, it can analyze failures using their complete context.
Jessica
You don’t need to copy and paste log lines or manually provide additional background information.
Jessica
Claude can retrieve the required context directly from the reports.
Jessica
You can also orchestrate and run collections of multiple test cases directly through the connection.
Jessica
Now, let’s compare the workflow with and without MCP.
Jessica
Without MCP, one task may require five separate tools—one for test generation, another for results, another for analysis and reporting, and another for device booking.
Jessica
Each activity is isolated in its own silo.
Jessica
Manually booking devices and tracking execution only addresses part of the problem because test execution is still triggered separately from Qpilot.
Jessica
MCP changes this by bringing the end-to-end workflow into the AI tool you’re already using.
Jessica
It also eliminates the need to debug failures one log line at a time.
Jessica
pCloudy’s AI failure analysis identifies the relevant issues for you.
Jessica
Claude can then provide a shareable report link and use the logs and failure details as context.
Jessica
These are some of the key advantages of using MCP to create and run tests, book devices, and manage manual or automated sessions.
Jessica
Even simple platform queries can otherwise take up time.
Jessica
For example, you can ask Claude to list available Android 15 devices, Google devices running Android 14 or 16, or any other configuration you need.
Jessica
It can return the available devices and the applications associated with your account.
Jessica
You can also ask for the latest build uploaded to pCloudy.
Jessica
This means you don’t need to log in every time you need basic information about the platform or its real devices.
Jessica
That saves time and helps you begin testing faster.
Jessica
Next, let’s look at our AI-powered failure analysis reports.
Jessica
These reports automatically transform raw logs into root-cause insights at both the test-case and build levels.
Jessica
They provide context for failures such as missing elements and performance problems.
Jessica
If a test experiences lag or latency, the report can identify possible causes.
Jessica
It also provides recommendations and suggested fixes.
Jessica
This AI-powered report is generated automatically for every test run.
Jessica
You receive analysis at both the build level and the individual test-case level.
Jessica
Let’s compare manual debugging with debugging through Claude and Qpilot MCP.
Jessica
In a manual workflow, you open the dashboard, locate a failed run, and cross-reference logs against screenshots or video.
Jessica
You then determine which video frame corresponds to each log line and develop a theory about the root cause.
Jessica
This can take a significant amount of time.
Jessica
You may also need to collaborate with multiple teams without a single source of truth.
Jessica
Debugging one issue can consume an entire afternoon or even a full day.
Jessica
With Claude and Qpilot MCP, Claude retrieves the failed-run logs and screenshots for you.
Jessica
Because pCloudy’s reporting already includes AI-generated root-cause analysis and the most relevant log lines, Claude can provide the context you need to investigate further.
Jessica
What matters is the relationship between the different pieces of evidence, rather than each source operating in isolation.
Jessica
Correlating that data manually is tedious and requires considerable switching between tools.
Jessica
This is where the connected workflow delivers a meaningful benefit.
Jessica
You can also generate a shareable summary and refer back to it later.
Jessica
If a test fails again, you can connect it with an earlier error and use the complete history to debug the issue more efficiently.
Jessica
With that, let’s move into the live demonstration of the Claude MCP solution.
Jessica
Before I demonstrate the MCP connection, I’ll briefly introduce pCloudy for anyone who isn’t familiar with our platform.
Jessica
pCloudy is a device cloud with intelligence and a SaaS-based platform that can be accessed from anywhere.
Jessica
At the core of our solution, we host real mobile devices and browsers.
Jessica
We operate data centers across multiple regions, allowing you to connect to devices, launch applications or URLs, and begin testing.
Jessica
In other words, we provide the test infrastructure.
Jessica
This is the landing page of the pCloudy platform.
Jessica
Every device listed here is a real, physical device hosted in one of our data-center locations.
Jessica
You can see the available locations, upload your builds, and start testing.
Jessica
The same process applies to our iOS devices.
Jessica
We also provide browser infrastructure with Windows and macOS machines.
Jessica
You can connect to different browsers and versions for cross-browser testing.
Jessica
The platform also supports integrations with popular automation frameworks.
Jessica
These include Appium, Selenium, Playwright, and several other tools listed in our documentation.
Jessica
Now, let’s look at Qpilot.
Jessica
Qpilot allows you to create automated test cases using plain-English descriptions.
Jessica
You can perform all these actions through the Qpilot interface.
Jessica
However, today I’ll focus on how you can create these tests without logging in to pCloudy.
Jessica
Qpilot is an AI-based test creation agent.
Jessica
You don’t need to learn special keywords; you simply describe the test in plain English.
Jessica
Qpilot converts that description into an end-to-end test case.
Jessica
It supports web, iOS, Android, and API testing on a single platform.
Jessica
Qpilot itself consolidates several tools.
Jessica
You might otherwise use one tool for API testing and another as a device farm.
Jessica
Qpilot combines pCloudy’s real-device infrastructure, an AI test creation agent, reporting, API testing, and integrations with Jira, Trello, and Slack.
Jessica
This already delivers significant tool consolidation, and MCP extends those benefits further.
Jessica
Now, let’s move to the Claude MCP demonstration.
Jessica
Here is my Claude account, and we provide documentation explaining how to enable MCP connections.
Jessica
In my list of connectors, you can see that the pCloudy toggle is already enabled.
Jessica
This means there is an active connection, and Claude can retrieve details from pCloudy and perform supported actions.
Jessica
The connection covers both pCloudy’s real-device infrastructure and Qpilot’s test creation capabilities.
Jessica
Let me begin with a simple request.
Jessica
I’ll ask Claude to list the available Google devices running Android versions 15 and 16.
Jessica
I’ve entered that request in plain English.
Jessica
Claude is now identifying the appropriate tools and retrieving the device list.
Jessica
These requests are being sent to pCloudy so Claude can obtain the relevant context.
Jessica
You can now see the available devices listed in the response.
Jessica
Similarly, I can ask Claude to list the available iOS devices.
Jessica
It retrieves the available configurations and operating-system versions for me.
Jessica
If you want to know whether a particular device is available or busy, or which devices can be used for automation, you can ask directly in Claude.
Jessica
You can even book a device through the same conversation.
Jessica
For example, I’ll ask Claude to connect to a Samsung device running Android 14 and install the IKEA app.
Jessica
Once I submit the instruction, Claude connects to the platform and begins performing the requested actions.
Jessica
A small amount of processing time is expected while Claude communicates with pCloudy.
Jessica
You can see each API call and action taking place behind the scenes while the device is being booked.
Jessica
Several individual actions are completed as part of this single request.
Jessica
I’ll return to pCloudy and check the active sessions.
Jessica
You can see that the platform has already selected the device and the connection is in progress.
Jessica
This should give you an idea of the interactions you can perform once the MCP connection is running.
Jessica
I’d also like to show how the connection works with the Qpilot agent.
Jessica
Here, I’ve asked Claude to create a new test case.
Jessica
I provided the test-case name, the APK or IPA name, and all the other information required to create it.
Jessica
I also specified that the test should be generated on a Google Pixel 2 XL.
Jessica
Claude is processing the request and completing the required action at each step.
Jessica
It first found the demo project, then located the appropriate folder and retrieved its folder ID.
Jessica
Once it has everything it needs, it creates the test case within that folder.
Jessica
The test case has now been created, and Claude has returned the session ID and a summary of the completed actions.
Jessica
Claude also suggests the next action and asks whether it should add more steps to the test case.
Jessica
I’ve responded with another plain-English description.
Jessica
You can see all seven steps being added to the test case.
Jessica
Claude has also summarized the outcome and confirmed that the test ran end to end on the Pixel 2 XL without any issues.
Jessica
You can also edit an existing test case.
Jessica
If you need to add a few steps, simply ask Claude to connect to the device and update the specified test case.
Jessica
Claude opens the test, adds the requested steps, and completes the update.
Jessica
This makes it easy to maintain and update existing test cases.
Jessica
On the Qpilot side, you can now see the test case with the exact name I specified in Claude.
Jessica
The automated test script has also been created.
Jessica
I can run it from either Claude or Qpilot.
Jessica
I can execute the test cases and retrieve the results through the same connected workflow.
Jessica
The test is now running, and I can monitor it live.
Jessica
Claude can provide links to the live execution and the final reports.
Jessica
Now, let’s look at the second part of the demonstration: rapid debugging with our AI failure analysis agent.
Jessica
Every automated test you run includes detailed reporting within the pCloudy platform.
Jessica
Here, you can see the most recent execution.
Jessica
When I open the build, I can view the status of every test case, including how many have completed, failed, timed out, or remain in progress.
Jessica
The platform also provides build-level analysis.
Jessica
It shows which sessions encountered an element-not-found error, which experienced slow page loads, and which devices had issues.
Jessica
Across past runs, you can identify devices with a higher tendency to fail and understand the errors associated with them.
Jessica
This allows you to debug at the build or project level.
Jessica
You can also open an individual test case for more detailed analysis.
Jessica
The text logs show the output of each step, including the clicks and actions performed.
Jessica
The step view shows every test step, the exact point of failure, and the corresponding error.
Jessica
If you need more detail, pCloudy still provides access to the complete raw and device logs.
Jessica
However, the AI analysis fast-tracks the investigation.
Jessica
In this example, the test failed because an element could not be found.
Jessica
The failure matches the error shown in the step view, and the analysis highlights the relevant log line.
Jessica
It identifies the exact log entry where the error occurred and recommends an action.
Jessica
In this case, the recommendation is to validate the element’s XPath.
Jessica
The report helps us pinpoint that the XPath used to select a location from the drop-down menu was probably incorrect.
Jessica
We can therefore identify the likely root cause without spending time searching through the logs manually.
Jessica
pCloudy also provides access to the complete video and screenshots.
Jessica
You can log bugs directly from the platform using integrations with Jira, Trello, Slack, and GitHub.
Jessica
The reports are shareable, which makes it easier to provide Claude with the relevant context.
Jessica
You can generate a shareable link and send it to your colleagues.
Jessica
They can access the full report even if they aren’t registered pCloudy users.
Jessica
That concludes this brief look at the failure analysis capabilities.
Jessica
That also brings us to the end of the live demonstration.
Jessica
I’ve shown how a real workflow can be managed through Claude, including retrieving results and reports.
Jessica
You can also ask Claude to analyze failures and view the entire workflow in one conversation.
Jessica
That workflow extends from booking a device and creating test cases through to execution, results, and analysis.
Jessica
Now, let’s look at how to get started.
Jessica
The setup is straightforward and takes only a few minutes.
Jessica
First, you need a pCloudy account and an AI tool such as Claude or Cursor.
Jessica
You then configure the Qpilot MCP server, connect it to your chosen tool, and begin generating tests and analyzing results.
Jessica
We provide complete documentation with all the required prerequisites and setup instructions.
Jessica
We’ll be happy to share it so you can connect MCP to your pCloudy account whenever you’re ready.
Jessica
Here, you can see the documentation for configuring the MCP server with pCloudy.
Jessica
It includes the prerequisites and exact steps required to complete the setup.
Jessica
I hope this gives you a clear understanding of what you can achieve and how you can optimize your workflow by connecting MCP with Qpilot and pCloudy.
Jessica
Please feel free to contact us if you have any questions or would like another demonstration or clarification.
Jessica
With that, we can move to the Q&A session.
Jessica
I don’t see any questions yet, so please add them using the Q&A tab.
Jessica
I can see that our first question has arrived.
Jessica
Do I need coding knowledge to use Claude with Qpilot MCP, or can a manual tester use it?
Jessica
Manual testers can definitely use it.
Jessica
Qpilot itself doesn’t require any coding knowledge.
Jessica
Once you follow the setup instructions and establish the MCP connection, you don’t need coding skills to use it.
Jessica
You can perform both automated and manual testing without writing code.
Jessica
How long does the setup take?
Jessica
It doesn’t take long; the setup requires only a few minutes.
Jessica
In less than half an hour, you can be up and running and create your first test script through Claude and MCP.
Jessica
Does this work with our existing test scripts, or do we have to start from scratch?
Jessica
You don’t need to start from scratch.
Jessica
If your scripts already exist in Qpilot, you can retrieve, edit, or execute them directly.
Jessica
Once MCP is enabled, it can also retrieve test cases that were created before the connection was configured.
Jessica
Does this replace our test automation framework, or does it work alongside it?
Jessica
If you use Qpilot, you can run your Qpilot test cases seamlessly through MCP.
Jessica
If you already have your own framework, pCloudy integrates with that as well.
Jessica
You can run Appium, Selenium, or Playwright test cases on our devices and browsers.
Jessica
The platform works whether you already have automation coverage and an established framework or need to create automation through Qpilot.
Jessica
I believe that covers all the questions.
Jessica
Please feel free to explore our Qpilot demonstrations or contact us if you would like a live demonstration.
Jessica
We’ll be happy to guide you through it.
Dinakar
Thank you so much, Jessica, for sharing those valuable insights into Claude, MCP, and how they work with Qpilot on the pCloudy platform.
Dinakar
We’d also like to thank everyone who took the time to join today’s webinar.
Dinakar
I’m sure some of Jessica’s ideas and insights have given you new possibilities to consider.
Dinakar
I recommend discussing these ideas with your teams and exploring how you can put them into practice.
00
00:00
Dinakar
Hello, everyone, and welcome to pCloudy’s webinar, “How QA Teams Are Saving Hours Every Sprint Using Claude with Qpilot MCP.”
Dinakar
My name is Dinakar, and I’ll be your host today.
Dinakar
Before we begin, I’d like to go over a few housekeeping points.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
If you have any questions or need clarification, please use the Q&A option at the bottom of your screen.
Dinakar
We’ll answer your questions during the Q&A session toward the end of the webinar.
Dinakar
Please also note that we’re recording this webinar.
Dinakar
We’ll share the recording afterward so you can watch it again or share it with your colleagues.
Dinakar
With that said, let me introduce today’s speaker, Jessica Selvin.
Dinakar
Jessica is Lead, Solution Engineering at pCloudy.
Dinakar
She has more than five years of experience in the IT industry.
Dinakar
Before joining pCloudy, she worked at LTIMindtree as a presales consultant, designing automation solutions for various industries.
Dinakar
She also has expertise in data analytics, machine learning, and artificial intelligence tools.
Dinakar
Without further ado, I’ll hand it over to Jessica.
Dinakar
Over to you, Jessica.
Jessica
Thank you so much for that introduction, Dinakar.
Jessica
I hope my screen is visible.
Jessica
As Dinakar mentioned, today’s topic is how Claude and the Qpilot MCP connection can help you generate tests faster, view all your test results in one place, and save hours otherwise spent switching between tools.
Jessica
Let’s begin by looking at what AI has already solved and why some testing activities still take so much time.
Jessica
Test generation, for example, has largely been solved by recent advances in AI.
Jessica
Whether you use an AI agent such as Qpilot or GitHub Copilot to write Playwright scripts, you can describe a flow and generate scripts quickly.
Jessica
That’s genuinely impressive, and the capability has become widely available very quickly.
Jessica
However, most QA teams still use several different tools.
Jessica
A QA engineer might use one tool to write a test, another to share the result, an AI assistant, an editor or IDE, a device farm, and separate dashboards and reports.
Jessica
They may then log bugs or tasks in Slack or Jira.
Jessica
This creates a significant amount of context switching.
Jessica
If API testing is involved, the engineer may also use a tool such as Postman.
Jessica
Debugging a single failure can require at least three or four manual handoffs.
Jessica
Each action and result must also be recorded separately in the relevant tool.
Jessica
This is where teams lose hours during every sprint or release cycle.
Jessica
As people switch between tools, important details are often lost, so their AI tools never receive the full context.
Jessica
Teams also spend time manually reviewing logs, dashboards, and reports.
Jessica
Meanwhile, much of their innovative work is happening inside AI tools such as Cursor or Claude Code.
Jessica
They can’t easily bring the context from other tools back into the environment they primarily use to develop scripts and generate new builds.
Jessica
That is the central problem the Qpilot and Claude MCP connection is designed to solve.
Jessica
Let me summarize the key agenda for today’s session.
Jessica
First, you’ll learn how Claude connects with Qpilot through MCP.
Jessica
You’ll see how to access pCloudy’s real devices directly from tools such as Cursor or Claude.
Jessica
Second, we’ll walk through a live demonstration.
Jessica
Third, I’ll cover our AI-powered test failure analysis reports and show how they significantly reduce debugging time.
Jessica
You’ll see how the system analyzes logs and provides detailed findings quickly.
Jessica
I’ll also explain how to execute test cases and create a reusable MCP setup for your team.
Jessica
As mentioned earlier, we’ll answer questions at the end of the demonstration.
Jessica
Let’s begin with the first section: where QA teams are still losing time and context between tools.
Dinakar
Hello, everyone, and welcome to pCloudy’s webinar, “How QA Teams Are Saving Hours Every Sprint Using Claude with Qpilot MCP.”
Dinakar
My name is Dinakar, and I’ll be your host today.
Dinakar
Before we begin, I’d like to go over a few housekeeping points.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
If you have any questions or need clarification, please use the Q&A option at the bottom of your screen.
Dinakar
We’ll answer your questions during the Q&A session toward the end of the webinar.
Dinakar
Please also note that we’re recording this webinar.
Dinakar
We’ll share the recording afterward so you can watch it again or share it with your colleagues.
Dinakar
With that said, let me introduce today’s speaker, Jessica Selvin.
Dinakar
Jessica is Lead, Solution Engineering at pCloudy.
Dinakar
She has more than five years of experience in the IT industry.
Dinakar
Before joining pCloudy, she worked at LTIMindtree as a presales consultant, designing automation solutions for various industries.
Dinakar
She also has expertise in data analytics, machine learning, and artificial intelligence tools.
Dinakar
Hello, everyone, and welcome to pCloudy’s webinar, “How QA Teams Are Saving Hours Every Sprint Using Claude with Qpilot MCP.”
Dinakar
My name is Dinakar, and I’ll be your host today.
Dinakar
Before we begin, I’d like to go over a few housekeeping points.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
If you have any questions or need clarification, please use the Q&A option at the bottom of your screen.
Dinakar
We’ll answer your questions during the Q&A session toward the end of the webinar.
Dinakar
Please also note that we’re recording this webinar.
Dinakar
We’ll share the recording afterward so you can watch it again or share it with your colleagues.
Dinakar
With that said, let me introduce today’s speaker, Jessica Selvin.
Dinakar
Jessica is Lead, Solution Engineering at pCloudy.
Dinakar
She has more than five years of experience in the IT industry.
Dinakar
Before joining pCloudy, she worked at LTIMindtree as a presales consultant, designing automation solutions for various indu00:00:00
Dinakar
Hello, everyone, and welcome to pCloudy’s webinar, “How QA Teams Are Saving Hours Every Sprint Using Claude with Qpilot MCP.”
Dinakar
My name is Dinakar, and I’ll be your host today.
Dinakar
Before we begin, I’d like to go over a few housekeeping points.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
If you have any questions or need clarification, please use the Q&A option at the bottom of your screen.
Dinakar
We’ll answer your questions during the Q&A session toward the end of the webinar.
Dinakar
Please also note that we’re recording this webinar.
Dinakar
We’ll share the recording afterward so you can watch it again or share it with your colleagues.
Dinakar
With that said, let me introduce today’s speaker, Jessica Selvin.
Dinakar
Jessica is Lead, Solution Engineering at pCloudy.
Dinakar
She has more than five years of experience in the IT industry.
Dinakar
Before joining pCloudy, she worked at LTIMindtree as a presales consultant, designing automation solutions for various industries.
Dinakar
She also has expertise in data analytics, machine learning, and artificial intelligence tools.
Dinakar
Without further ado, I’ll hand it over to Jessica.
Dinakar
Over to you, Jessica.
Jessica
Thank you so much for that introduction, Dinakar.
Jessica
I hope my screen is visible.
Jessica
As Dinakar mentioned, today’s topic is how Claude and the Qpilot MCP connection can help you generate tests faster, view all your test results in one place, and save hours otherwise spent switching between tools.
Jessica
Let’s begin by looking at what AI has already solved and why some testing activities still take so much time.
Jessica
Test generation, for example, has largely been solved by recent advances in AI.
Jessica
Whether you use an AI agent such as Qpilot or GitHub Copilot to write Playwright scripts, you can describe a flow and generate scripts quickly.
Jessica
That’s genuinely impressive, and the capability has become widely available very quickly.
Jessica
However, most QA teams still use several different tools.
Jessica
A QA engineer might use one tool to write a test, another to share the result, an AI assistant, an editor or IDE, a device farm, and separate dashboards and reports.
Jessica
They may then log bugs or tasks in Slack or Jira.
Jessica
This creates a significant amount of context switching.
Jessica
If API testing is involved, the engineer may also use a tool such as Postman.
Jessica
Debugging a single failure can require at least three or four manual handoffs.
Jessica
Each action and result must also be recorded separately in the relevant tool.
Jessica
This is where teams lose hours during every sprint or release cycle.
Jessica
As people switch between tools, important details are often lost, so their AI tools never receive the full context.
Jessica
Teams also spend time manually reviewing logs, dashboards, and reports.
Jessica
Meanwhile, much of their innovative work is happening inside AI tools such as Cursor or Claude Code.
Jessica
They can’t easily bring the context from other tools back into the environment they primarily use to develop scripts and generate new builds.
Jessica
That is the central problem the Qpilot and Claude MCP connection is designed to solve.
Jessica
Let me summarize the key agenda for today’s session.
Jessica
First, you’ll learn how Claude connects with Qpilot through MCP.
Jessica
You’ll see how to access pCloudy’s real devices directly from tools such as Cursor or Claude.
Jessica
Second, we’ll walk through a live demonstration.
Jessica
Third, I’ll cover our AI-powered test failure analysis reports and show how they significantly reduce debugging time.
Jessica
You’ll see how the system analyzes logs and provides detailed findings quickly.
Jessica
I’ll also explain how to execute test cases and create a reusable MCP setup for your team.
Jessica
As mentioned earlier, we’ll answer questions at the end of the demonstration.
Jessica
Let’s begin with the first section: where QA teams are still losing time and context between tools.
Jessica
In a typical workflow, you often begin with an AI assistant such as Cursor, GitHub Copilot, or Claude Code.
Jessica
You describe the flow you want to test, refine the output through a few exchanges, and eventually receive a script that is ready to run.
Jessica
Next, you copy the script from the chat window and paste it into your codebase.
Jessica
You then fix imports, apply page-object conventions, and replace locators where necessary.
Jessica
This can create further back-and-forth work in your IDE or test editor.
Jessica
When it’s time to use a device farm, you must book a device, add the required capabilities, connect to the device, and run your test cases.
Jessica
You may switch between your IDE or CI/CD tool and a device farm such as pCloudy to track execution.
Jessica
You may also open a separate dashboarding solution to review the results.
Jessica
Finally, you update documentation or log bugs in tools such as Slack or Jira.
Jessica
In this workflow, you’ve switched between five tools and created five opportunities to lose context.
Jessica
You may lose critical details or fixes made in the test editor and fail to bring that context back into your AI assistant.
Jessica
This limits your ability to improve the accuracy of the scripts it generates.
Jessica
That lost context eventually translates into a significant loss of time.
Jessica
Context switching is only one part of the problem.
Jessica
Teams also spend time booking devices, uploading builds, and tracking execution status.
Jessica
Each time, you may need to open a dashboard or active session to check whether execution has finished and whether the report has been generated.
Jessica
These may seem like simple tasks, but they accumulate and consume a meaningful amount of time.
Jessica
Another time-consuming task is reviewing raw logs.
Jessica
When a test case fails, you may need to examine extensive logs to understand what went wrong.
Jessica
With pCloudy’s AI failure analysis agent, you can immediately see the relevant log lines and identify where the test failed.
Jessica
Reporting can create another delay, especially when test results are stored in a separate tool.
Jessica
You still need to retrieve the report, debug the issue, and send the findings back to Jira or Slack.
Jessica
These are the main touchpoints where teams lose both time and context.
Jessica
Now, let me introduce the Claude and Qpilot MCP connection.
Jessica
Claude is the focus of today’s demonstration, but pCloudy also integrates with tools such as Cursor.
Jessica
MCP stands for Model Context Protocol.
Jessica
It attracted considerable attention a few months ago and is an open standard introduced by Anthropic.
Jessica
Put simply, MCP allows an AI assistant to communicate directly with an external system instead of requiring you to act as the intermediary.
Jessica
It enables the assistant to retrieve data from your tools and perform actions on your platform.
Jessica
That is the core idea behind an MCP connection.
Jessica
It provides a common protocol for interacting with different tools and simplifies those connections.
Jessica
You can think of it as an API-like connection between Claude and pCloudy.
Jessica
Instead of reading information in one system, retyping it into another, and carrying the context in your head, MCP connects the systems directly.
Jessica
It removes you from the relay loop and lets your AI assistant communicate directly with the platform.
Jessica
If you haven’t heard of Qpilot, it is pCloudy’s AI-based test creation solution.
Jessica
You can explore our other Qpilot webinars, watch demonstrations on our website, or contact us to learn more.
Jessica
Qpilot is an AI-powered test creation platform built on top of pCloudy’s real-device infrastructure.
Jessica
pCloudy provides real Android and iOS devices, as well as a browser cloud, for manual and automated mobile and web testing.
Jessica
You can run tests using Appium, Selenium, Playwright, and other popular frameworks.
Jessica
Alternatively, you can use Qpilot, our AI layer, to convert plain-English descriptions into automated test steps.
Jessica
Real-device infrastructure is the foundation of the pCloudy platform.
Jessica
We host physical devices, not simulators or emulators, in our device cloud.
Jessica
Through Claude, you can access both Qpilot’s AI capabilities and pCloudy’s real-device infrastructure.
Jessica
Let’s look at some of the practical tasks this connection can support.
Jessica
First, you can generate tests directly from Claude instead of opening Qpilot separately.
Jessica
Once the pCloudy MCP connection is enabled, you can describe the test you need, ask Claude to create it, or retrieve an existing test case.
Jessica
You can also retrieve results and reports, and monitor live executions.
Jessica
This applies to both automated and manual sessions.
Jessica
Because Claude can access those reports, it can analyze failures using their complete context.
Jessica
You don’t need to copy and paste log lines or manually provide additional background information.
Jessica
Claude can retrieve the required context directly from the reports.
Jessica
You can also orchestrate and run collections of multiple test cases directly through the connection.
Jessica
Now, let’s compare the workflow with and without MCP.
Jessica
Without MCP, one task may require five separate tools—one for test generation, another for results, another for analysis and reporting, and another for device booking.
Jessica
Each activity is isolated in its own silo.
Jessica
Manually booking devices and tracking execution only addresses part of the problem because test execution is still triggered separately from Qpilot.
Jessica
MCP changes this by bringing the end-to-end workflow into the AI tool you’re already using.
Jessica
It also eliminates the need to debug failures one log line at a time.
Jessica
pCloudy’s AI failure analysis identifies the relevant issues for you.
Jessica
Claude can then provide a shareable report link and use the logs and failure details as context.
Jessica
These are some of the key advantages of using MCP to create and run tests, book devices, and manage manual or automated sessions.
Jessica
Even simple platform queries can otherwise take up time.
Jessica
For example, you can ask Claude to list available Android 15 devices, Google devices running Android 14 or 16, or any other configuration you need.
Jessica
It can return the available devices and the applications associated with your account.
Jessica
You can also ask for the latest build uploaded to pCloudy.
Jessica
This means you don’t need to log in every time you need basic information about the platform or its real devices.
Jessica
That saves time and helps you begin testing faster.
Jessica
Next, let’s look at our AI-powered failure analysis reports.
Jessica
These reports automatically transform raw logs into root-cause insights at both the test-case and build levels.
Jessica
They provide context for failures such as missing elements and performance problems.
Jessica
If a test experiences lag or latency, the report can identify possible causes.
Jessica
It also provides recommendations and suggested fixes.
Jessica
This AI-powered report is generated automatically for every test run.
Jessica
You receive analysis at both the build level and the individual test-case level.
Jessica
Let’s compare manual debugging with debugging through Claude and Qpilot MCP.
Jessica
In a manual workflow, you open the dashboard, locate a failed run, and cross-reference logs against screenshots or video.
Jessica
You then determine which video frame corresponds to each log line and develop a theory about the root cause.
Jessica
This can take a significant amount of time.
Jessica
You may also need to collaborate with multiple teams without a single source of truth.
Jessica
Debugging one issue can consume an entire afternoon or even a full day.
Jessica
With Claude and Qpilot MCP, Claude retrieves the failed-run logs and screenshots for you.
Jessica
Because pCloudy’s reporting already includes AI-generated root-cause analysis and the most relevant log lines, Claude can provide the context you need to investigate further.
Jessica
What matters is the relationship between the different pieces of evidence, rather than each source operating in isolation.
Jessica
Correlating that data manually is tedious and requires considerable switching between tools.
Jessica
This is where the connected workflow delivers a meaningful benefit.
Jessica
You can also generate a shareable summary and refer back to it later.
Jessica
If a test fails again, you can connect it with an earlier error and use the complete history to debug the issue more efficiently.
Jessica
With that, let’s move into the live demonstration of the Claude MCP solution.
Jessica
Before I demonstrate the MCP connection, I’ll briefly introduce pCloudy for anyone who isn’t familiar with our platform.
Jessica
pCloudy is a device cloud with intelligence and a SaaS-based platform that can be accessed from anywhere.
Jessica
At the core of our solution, we host real mobile devices and browsers.
Jessica
We operate data centers across multiple regions, allowing you to connect to devices, launch applications or URLs, and begin testing.
Jessica
In other words, we provide the test infrastructure.
Jessica
This is the landing page of the pCloudy platform.
Jessica
Every device listed here is a real, physical device hosted in one of our data-center locations.
Jessica
You can see the available locations, upload your builds, and start testing.
Jessica
The same process applies to our iOS devices.
Jessica
We also provide browser infrastructure with Windows and macOS machines.
Jessica
You can connect to different browsers and versions for cross-browser testing.
Jessica
The platform also supports integrations with popular automation frameworks.
Jessica
These include Appium, Selenium, Playwright, and several other tools listed in our documentation.
Jessica
Now, let’s look at Qpilot.
Jessica
Qpilot allows you to create automated test cases using plain-English descriptions.
Jessica
You can perform all these actions through the Qpilot interface.
Jessica
However, today I’ll focus on how you can create these tests without logging in to pCloudy.
Jessica
Qpilot is an AI-based test creation agent.
Jessica
You don’t need to learn special keywords; you simply describe the test in plain English.
Jessica
Qpilot converts that description into an end-to-end test case.
Jessica
It supports web, iOS, Android, and API testing on a single platform.
Jessica
Qpilot itself consolidates several tools.
Jessica
You might otherwise use one tool for API testing and another as a device farm.
Jessica
Qpilot combines pCloudy’s real-device infrastructure, an AI test creation agent, reporting, API testing, and integrations with Jira, Trello, and Slack.
Jessica
This already delivers significant tool consolidation, and MCP extends those benefits further.
Jessica
Now, let’s move to the Claude MCP demonstration.
Jessica
Here is my Claude account, and we provide documentation explaining how to enable MCP connections.
Jessica
In my list of connectors, you can see that the pCloudy toggle is already enabled.
Jessica
This means there is an active connection, and Claude can retrieve details from pCloudy and perform supported actions.
Jessica
The connection covers both pCloudy’s real-device infrastructure and Qpilot’s test creation capabilities.
Jessica
Let me begin with a simple request.
Jessica
I’ll ask Claude to list the available Google devices running Android versions 15 and 16.
Jessica
I’ve entered that request in plain English.
Jessica
Claude is now identifying the appropriate tools and retrieving the device list.
Jessica
These requests are being sent to pCloudy so Claude can obtain the relevant context.
Jessica
You can now see the available devices listed in the response.
Jessica
Similarly, I can ask Claude to list the available iOS devices.
Jessica
It retrieves the available configurations and operating-system versions for me.
Jessica
If you want to know whether a particular device is available or busy, or which devices can be used for automation, you can ask directly in Claude.
Jessica
You can even book a device through the same conversation.
Jessica
For example, I’ll ask Claude to connect to a Samsung device running Android 14 and install the IKEA app.
Jessica
Once I submit the instruction, Claude connects to the platform and begins performing the requested actions.
Jessica
A small amount of processing time is expected while Claude communicates with pCloudy.
Jessica
You can see each API call and action taking place behind the scenes while the device is being booked.
Jessica
Several individual actions are completed as part of this single request.
Jessica
I’ll return to pCloudy and check the active sessions.
Jessica
You can see that the platform has already selected the device and the connection is in progress.
Jessica
This should give you an idea of the interactions you can perform once the MCP connection is running.
Jessica
I’d also like to show how the connection works with the Qpilot agent.
Jessica
Here, I’ve asked Claude to create a new test case.
Jessica
I provided the test-case name, the APK or IPA name, and all the other information required to create it.
Jessica
I also specified that the test should be generated on a Google Pixel 2 XL.
Jessica
Claude is processing the request and completing the required action at each step.
Jessica
It first found the demo project, then located the appropriate folder and retrieved its folder ID.
Jessica
Once it has everything it needs, it creates the test case within that folder.
Jessica
The test case has now been created, and Claude has returned the session ID and a summary of the completed actions.
Jessica
Claude also suggests the next action and asks whether it should add more steps to the test case.
Jessica
I’ve responded with another plain-English description.
Jessica
You can see all seven steps being added to the test case.
Jessica
Claude has also summarized the outcome and confirmed that the test ran end to end on the Pixel 2 XL without any issues.
Jessica
You can also edit an existing test case.
Jessica
If you need to add a few steps, simply ask Claude to connect to the device and update the specified test case.
Jessica
Claude opens the test, adds the requested steps, and completes the update.
Jessica
This makes it easy to maintain and update existing test cases.
Jessica
On the Qpilot side, you can now see the test case with the exact name I specified in Claude.
Jessica
The automated test script has also been created.
Jessica
I can run it from either Claude or Qpilot.
Jessica
I can execute the test cases and retrieve the results through the same connected workflow.
Jessica
The test is now running, and I can monitor it live.
Jessica
Claude can provide links to the live execution and the final reports.
Jessica
Now, let’s look at the second part of the demonstration: rapid debugging with our AI failure analysis agent.
Jessica
Every automated test you run includes detailed reporting within the pCloudy platform.
Jessica
Here, you can see the most recent execution.
Jessica
When I open the build, I can view the status of every test case, including how many have completed, failed, timed out, or remain in progress.
Jessica
The platform also provides build-level analysis.
Jessica
It shows which sessions encountered an element-not-found error, which experienced slow page loads, and which devices had issues.
Jessica
Across past runs, you can identify devices with a higher tendency to fail and understand the errors associated with them.
Jessica
This allows you to debug at the build or project level.
Jessica
You can also open an individual test case for more detailed analysis.
Jessica
The text logs show the output of each step, including the clicks and actions performed.
Jessica
The step view shows every test step, the exact point of failure, and the corresponding error.
Jessica
If you need more detail, pCloudy still provides access to the complete raw and device logs.
Jessica
However, the AI analysis fast-tracks the investigation.
Jessica
In this example, the test failed because an element could not be found.
Jessica
The failure matches the error shown in the step view, and the analysis highlights the relevant log line.
Jessica
It identifies the exact log entry where the error occurred and recommends an action.
Jessica
In this case, the recommendation is to validate the element’s XPath.
Jessica
The report helps us pinpoint that the XPath used to select a location from the drop-down menu was probably incorrect.
Jessica
We can therefore identify the likely root cause without spending time searching through the logs manually.
Jessica
pCloudy also provides access to the complete video and screenshots.
Jessica
You can log bugs directly from the platform using integrations with Jira, Trello, Slack, and GitHub.
Jessica
The reports are shareable, which makes it easier to provide Claude with the relevant context.
Jessica
You can generate a shareable link and send it to your colleagues.
Jessica
They can access the full report even if they aren’t registered pCloudy users.
Jessica
That concludes this brief look at the failure analysis capabilities.
Jessica
That also brings us to the end of the live demonstration.
Jessica
I’ve shown how a real workflow can be managed through Claude, including retrieving results and reports.
Jessica
You can also ask Claude to analyze failures and view the entire workflow in one conversation.
Jessica
That workflow extends from booking a device and creating test cases through to execution, results, and analysis.
Jessica
Now, let’s look at how to get started.
Jessica
The setup is straightforward and takes only a few minutes.
Jessica
First, you need a pCloudy account and an AI tool such as Claude or Cursor.
Jessica
You then configure the Qpilot MCP server, connect it to your chosen tool, and begin generating tests and analyzing results.
Jessica
We provide complete documentation with all the required prerequisites and setup instructions.
Jessica
We’ll be happy to share it so you can connect MCP to your pCloudy account whenever you’re ready.
Jessica
Here, you can see the documentation for configuring the MCP server with pCloudy.
Jessica
It includes the prerequisites and exact steps required to complete the setup.
Jessica
I hope this gives you a clear understanding of what you can achieve and how you can optimize your workflow by connecting MCP with Qpilot and pCloudy.
Jessica
Please feel free to contact us if you have any questions or would like another demonstration or clarification.
Jessica
With that, we can move to the Q&A session.
Jessica
I don’t see any questions yet, so please add them using the Q&A tab.
Jessica
I can see that our first question has arrived.
Jessica
Do I need coding knowledge to use Claude with Qpilot MCP, or can a manual tester use it?
Jessica
Manual testers can definitely use it.
Jessica
Qpilot itself doesn’t require any coding knowledge.
Jessica
Once you follow the setup instructions and establish the MCP connection, you don’t need coding skills to use it.
Jessica
You can perform both automated and manual testing without writing code.
Jessica
How long does the setup take?
Jessica
It doesn’t take long; the setup requires only a few minutes.
Jessica
In less than half an hour, you can be up and running and create your first test script through Claude and MCP.
Jessica
Does this work with our existing test scripts, or do we have to start from scratch?
Jessica
You don’t need to start from scratch.
stries.
Dinakar
She also has expertise in data analytics, machine learning, and artificial intelligence tools.
Dinakar
Without further ado, I’ll hand it over to Jessica.
Dinakar
Over to you, Jessica.
Jessica
Thank you so much for that introduction, Dinakar.
Jessica
I hope my screen is visible.
Jessica
As Dinakar mentioned, today’s topic is how Claude and the Qpilot MCP connection can help you generate tests faster, view all your test results in one place, and save hours otherwise spent switching between tools.
Jessica
Let’s begin by looking at what AI has already solved and why some testing activities still take so much time.
Jessica
Test generation, for example, has largely been solved by recent advances in AI.
Jessica
Whether you use an AI agent such as Qpilot or GitHub Copilot to write Playwright scripts, you can describe a flow and generate scripts quickly.
Jessica
That’s genuinely impressive, and the capability has become widely available very quickly.
Jessica
However, most QA teams still use several different tools.
Jessica
A QA engineer might use one tool to write a test, another to share the result, an AI assistant, an editor or IDE, a device farm, and separate dashboards and reports.
Jessica
They may then log bugs or tasks in Slack or Jira.
Jessica
This creates a significant amount of context switching.
Jessica
If API testing is involved, the engineer may also use a tool such as Postman.
Jessica
Debugging a single failure can require at least three or four manual handoffs.
Jessica
Each action and result must also be recorded separately in the relevant tool.
Jessica
This is where teams lose hours during every sprint or release cycle.
Jessica
As people switch between tools, important details are often lost, so their AI tools never receive the full context.
Jessica
Teams also spend time manually reviewing logs, dashboards, and reports.
Jessica
Meanwhile, much of their innovative work is happening inside AI tools such as Cursor or Claude Code.
Jessica
They can’t easily bring the context from other tools back into the environment they primarily use to develop scripts and generate new builds.
Jessica
That is the central problem the Qpilot and Claude MCP connection is designed to solve.
Jessica
Let me summarize the key agenda for today’s session.
Jessica
First, you’ll learn how Claude connects with Qpilot through MCP.
Jessica
You’ll see how to access pCloudy’s real devices directly from tools such as Cursor or Claude.
Jessica
Second, we’ll walk through a live demonstration.
Jessica
Third, I’ll cover our AI-powered test failure analysis reports and show how they significantly reduce debugging time.
Jessica
You’ll see how the system analyzes logs and provides detailed findings quickly.
Jessica
I’ll also explain how to execute test cases and create a reusable MCP setup for your team.
Jessica
As mentioned earlier, we’ll answer questions at the end of the demonstration.
Jessica
Let’s begin with the first section: where QA teams are still losing time and context between tools.
Jessica
In a typical workflow, you often begin with an AI assistant such as Cursor, GitHub Copilot, or Claude Code.
Jessica
You describe the flow you want to test, refine the output through a few exchanges, and eventually receive a script that is ready to run.
Jessica
Next, you copy the script from the chat window and paste it into your codebase.
Jessica
You then fix imports, apply page-object conventions, and replace locators where necessary.
Jessica
This can create further back-and-forth work in your IDE or test editor.
Jessica
When it’s time to use a device farm, you must book a device, add the required capabilities, connect to the device, and run your test cases.
Jessica
You may switch between your IDE or CI/CD tool and a device farm such as pCloudy to track execution.
Jessica
You may also open a separate dashboarding solution to review the results.
Jessica
Finally, you update documentation or log bugs in tools such as Slack or Jira.
Jessica
In this workflow, you’ve switched between five tools and created five opportunities to lose context.
Jessica
You may lose critical details or fixes made in the test editor and fail to bring that context back into your AI assistant.
Jessica
This limits your ability to improve the accuracy of the scripts it generates.
Jessica
That lost context eventually translates into a significant loss of time.
Jessica
Context switching is only one part of the problem.
Jessica
Teams also spend time booking devices, uploading builds, and tracking execution status.
Jessica
Each time, you may need to open a dashboard or active session to check whether execution has finished and whether the report has been generated.
Jessica
These may seem like simple tasks, but they accumulate and consume a meaningful amount of time.
Jessica
Another time-consuming task is reviewing raw logs.
Jessica
When a test case fails, you may need to examine extensive logs to understand what went wrong.
Jessica
With pCloudy’s AI failure analysis agent, you can immediately see the relevant log lines and identify where the test failed.
Jessica
Reporting can create another delay, especially when test results are stored in a separate tool.
Jessica
You still need to retrieve the report, debug the issue, and send the findings back to Jira or Slack.
Jessica
These are the main touchpoints where teams lose both time and context.
Jessica
Now, let me introduce the Claude and Qpilot MCP connection.
Jessica
Claude is the focus of today’s demonstration, but pCloudy also integrates with tools such as Cursor.
Jessica
MCP stands for Model Context Protocol.
Jessica
It attracted considerable attention a few months ago and is an open standard introduced by Anthropic.
Jessica
Put simply, MCP allows an AI assistant to communicate directly with an external system instead of requiring you to act as the intermediary.
Jessica
It enables the assistant to retrieve data from your tools and perform actions on your platform.
Jessica
That is the core idea behind an MCP connection.
Jessica
It provides a common protocol for interacting with different tools and simplifies those connections.
Jessica
You can think of it as an API-like connection between Claude and pCloudy.
Jessica
Instead of reading information in one system, retyping it into another, and carrying the context in your head, MCP connects the systems directly.
Jessica
It removes you from the relay loop and lets your AI assistant communicate directly with the platform.
Jessica
If you haven’t heard of Qpilot, it is pCloudy’s AI-based test creation solution.
Jessica
You can explore our other Qpilot webinars, watch demonstrations on our website, or contact us to learn more.
Jessica
Qpilot is an AI-powered test creation platform built on top of pCloudy’s real-device infrastructure.
Jessica
pCloudy provides real Android and iOS devices, as well as a browser cloud, for manual and automated mobile and web testing.
Jessica
You can run tests using Appium, Selenium, Playwright, and other popular frameworks.
Jessica
Alternatively, you can use Qpilot, our AI layer, to convert plain-English descriptions into automated test steps.
Jessica
Real-device infrastructure is the foundation of the pCloudy platform.
Jessica
We host physical devices, not simulators or emulators, in our device cloud.
Jessica
Through Claude, you can access both Qpilot’s AI capabilities and pCloudy’s real-device infrastructure.
Jessica
Let’s look at some of the practical tasks this connection can support.
Jessica
First, you can generate tests directly from Claude instead of opening Qpilot separately.
Jessica
Once the pCloudy MCP connection is enabled, you can describe the test you need, ask Claude to create it, or retrieve an existing test case.
Jessica
You can also retrieve results and reports, and monitor live executions.
Jessica
This applies to both automated and manual sessions.
Jessica
Because Claude can access those reports, it can analyze failures using their complete context.
Jessica
You don’t need to copy and paste log lines or manually provide additional background information.
Jessica
Claude can retrieve the required context directly from the reports.
Jessica
You can also orchestrate and run collections of multiple test cases directly through the connection.
Jessica
Now, let’s compare the workflow with and without MCP.
Jessica
Without MCP, one task may require five separate tools—one for test generation, another for results, another for analysis and reporting, and another for device booking.
Jessica
Each activity is isolated in its own silo.
Jessica
Manually booking devices and tracking execution only addresses part of the problem because test execution is still triggered separately from Qpilot.
Jessica
MCP changes this by bringing the end-to-end workflow into the AI tool you’re already using.
Jessica
It also eliminates the need to debug failures one log line at a time.
Jessica
pCloudy’s AI failure analysis identifies the relevant issues for you.
Jessica
Claude can then provide a shareable report link and use the logs and failure details as context.
Jessica
These are some of the key advantages of using MCP to create and run tests, book devices, and manage manual or automated sessions.
Jessica
Even simple platform queries can otherwise take up time.
Jessica
For example, you can ask Claude to list available Android 15 devices, Google devices running Android 14 or 16, or any other configuration you need.
Jessica
It can return the available devices and the applications associated with your account.
Jessica
You can also ask for the latest build uploaded to pCloudy.
Jessica
This means you don’t need to log in every time you need basic information about the platform or its real devices.
Jessica
That saves time and helps you begin testing faster.
Jessica
Next, let’s look at our AI-powered failure analysis reports.
Jessica
These reports automatically transform raw logs into root-cause insights at both the test-case and build levels.
Jessica
They provide context for failures such as missing elements and performance problems.
Jessica
If a test experiences lag or latency, the report can identify possible causes.
Jessica
It also provides recommendations and suggested fixes.
Jessica
This AI-powered report is generated automatically for every test run.
Jessica
You receive analysis at both the build level and the individual test-case level.
Jessica
Let’s compare manual debugging with debugging through Claude and Qpilot MCP.
Jessica
In a manual workflow, you open the dashboard, locate a failed run, and cross-reference logs against screenshots or video.
Jessica
You then determine which video frame corresponds to each log line and develop a theory about the root cause.
Jessica
This can take a significant amount of time.
Jessica
You may also need to collaborate with multiple teams without a single source of truth.
Jessica
Debugging one issue can consume an entire afternoon or even a full day.
Jessica
With Claude and Qpilot MCP, Claude retrieves the failed-run logs and screenshots for you.
Jessica
Because pCloudy’s reporting already includes AI-generated root-cause analysis and the most relevant log lines, Claude can provide the context you need to investigate further.
Jessica
What matters is the relationship between the different pieces of evidence, rather than each source operating in isolation.
Jessica
Correlating that data manually is tedious and requires considerable switching between tools.
Jessica
This is where the connected workflow delivers a meaningful benefit.
Jessica
You can also generate a shareable summary and refer back to it later.
Jessica
If a test fails again, you can connect it with an earlier error and use the complete history to debug the issue more efficiently.
Jessica
With that, let’s move into the live demonstration of the Claude MCP solution.
Jessica
Before I demonstrate the MCP connection, I’ll briefly introduce pCloudy for anyone who isn’t familiar with our platform.
Jessica
pCloudy is a device cloud with intelligence and a SaaS-based platform that can be accessed from anywhere.
Jessica
At the core of our solution, we host real mobile devices and browsers.
Jessica
We operate data centers across multiple regions, allowing you to connect to devices, launch applications or URLs, and begin testing.
Jessica
In other words, we provide the test infrastructure.
Jessica
This is the landing page of the pCloudy platform.
Jessica
Every device listed here is a real, physical device hosted in one of our data-center locations.
Jessica
You can see the available locations, upload your builds, and start testing.
Jessica
The same process applies to our iOS devices.
Jessica
We also provide browser infrastructure with Windows and macOS machines.
Jessica
You can connect to different browsers and versions for cross-browser testing.
Jessica
The platform also supports integrations with popular automation frameworks.
Jessica
These include Appium, Selenium, Playwright, and several other tools listed in our documentation.
Jessica
Now, let’s look at Qpilot.
Jessica
Qpilot allows you to create automated test cases using plain-English descriptions.
Jessica
You can perform all these actions through the Qpilot interface.
Jessica
However, today I’ll focus on how you can create these tests without logging in to pCloudy.
Jessica
Qpilot is an AI-based test creation agent.
Jessica
You don’t need to learn special keywords; you simply describe the test in plain English.
Jessica
Qpilot converts that description into an end-to-end test case.
Jessica
It supports web, iOS, Android, and API testing on a single platform.
Jessica
Qpilot itself consolidates several tools.
Jessica
You might otherwise use one tool for API testing and another as a device farm.
Jessica
Qpilot combines pCloudy’s real-device infrastructure, an AI test creation agent, reporting, API testing, and integrations with Jira, Trello, and Slack.
Jessica
This already delivers significant tool consolidation, and MCP extends those benefits further.
Jessica
Now, let’s move to the Claude MCP demonstration.
Jessica
Here is my Claude account, and we provide documentation explaining how to enable MCP connections.
Jessica
In my list of connectors, you can see that the pCloudy toggle is already enabled.
Jessica
This means there is an active connection, and Claude can retrieve details from pCloudy and perform supported actions.
Jessica
The connection covers both pCloudy’s real-device infrastructure and Qpilot’s test creation capabilities.
Jessica
Let me begin with a simple request.
Jessica
I’ll ask Claude to list the available Google devices running Android versions 15 and 16.
Jessica
I’ve entered that request in plain English.
Jessica
Claude is now identifying the appropriate tools and retrieving the device list.
Jessica
These requests are being sent to pCloudy so Claude can obtain the relevant context.
Jessica
You can now see the available devices listed in the response.
Jessica
Similarly, I can ask Claude to list the available iOS devices.
Jessica
It retrieves the available configurations and operating-system versions for me.
Jessica
If you want to know whether a particular device is available or busy, or which devices can be used for automation, you can ask directly in Claude.
Jessica
You can even book a device through the same conversation.
Jessica
For example, I’ll ask Claude to connect to a Samsung device running Android 14 and install the IKEA app.
Jessica
Once I submit the instruction, Claude connects to the platform and begins performing the requested actions.
Jessica
A small amount of processing time is expected while Claude communicates with pCloudy.
Jessica
You can see each API call and action taking place behind the scenes while the device is being booked.
Jessica
Several individual actions are completed as part of this single request.
Jessica
I’ll return to pCloudy and check the active sessions.
Jessica
You can see that the platform has already selected the device and the connection is in progress.
Jessica
This should give you an idea of the interactions you can perform once the MCP connection is running.
Jessica
I’d also like to show how the connection works with the Qpilot agent.
Jessica
Here, I’ve asked Claude to create a new test case.
Jessica
I provided the test-case name, the APK or IPA name, and all the other information required to create it.
Jessica
I also specified that the test should be generated on a Google Pixel 2 XL.
Jessica
Claude is processing the request and completing the required action at each step.
Jessica
It first found the demo project, then located the appropriate folder and retrieved its folder ID.
Jessica
Once it has everything it needs, it creates the test case within that folder.
Jessica
The test case has now been created, and Claude has returned the session ID and a summary of the completed actions.
Jessica
Claude also suggests the next action and asks whether it should add more steps to the test case.
Jessica
I’ve responded with another plain-English description.
Jessica
You can see all seven steps being added to the test case.
Jessica
Claude has also summarized the outcome and confirmed that the test ran end to end on the Pixel 2 XL without any issues.
Jessica
You can also edit an existing test case.
Jessica
If you need to add a few steps, simply ask Claude to connect to the device and update the specified test case.
Jessica
Claude opens the test, adds the requested steps, and completes the update.
Jessica
This makes it easy to maintain and update existing test cases.
Jessica
On the Qpilot side, you can now see the test case with the exact name I specified in Claude.
Jessica
The automated test script has also been created.
Jessica
I can run it from either Claude or Qpilot.
Jessica
I can execute the test cases and retrieve the results through the same connected workflow.
Jessica
The test is now running, and I can monitor it live.
Jessica
Claude can provide links to the live execution and the final reports.
Jessica
Now, let’s look at the second part of the demonstration: rapid debugging with our AI failure analysis agent.
Jessica
Every automated test you run includes detailed reporting within the pCloudy platform.
Jessica
Here, you can see the most recent execution.
Jessica
When I open the build, I can view the status of every test case, including how many have completed, failed, timed out, or remain in progress.
Jessica
The platform also provides build-level analysis.
Jessica
It shows which sessions encountered an element-not-found error, which experienced slow page loads, and which devices had issues.
Jessica
Across past runs, you can identify devices with a higher tendency to fail and understand the errors associated with them.
Jessica
This allows you to debug at the build or project level.
Jessica
You can also open an individual test case for more detailed analysis.
Jessica
The text logs show the output of each step, including the clicks and actions performed.
Jessica
The step view shows every test step, the exact point of failure, and the corresponding error.
Jessica
If you need more detail, pCloudy still provides access to the complete raw and device logs.
Jessica
However, the AI analysis fast-tracks the investigation.
Jessica
In this example, the test failed because an element could not be found.
Jessica
The failure matches the error shown in the step view, and the analysis highlights the relevant log line.
Jessica
It identifies the exact log entry where the error occurred and recommends an action.
Jessica
In this case, the recommendation is to validate the element’s XPath.
Jessica
The report helps us pinpoint that the XPath used to select a location from the drop-down menu was probably incorrect.
Jessica
We can therefore identify the likely root cause without spending time searching through the logs manually.
Jessica
pCloudy also provides access to the complete video and screenshots.
Jessica
You can log bugs directly from the platform using integrations with Jira, Trello, Slack, and GitHub.
Jessica
The reports are shareable, which makes it easier to provide Claude with the relevant context.
Jessica
You can generate a shareable link and send it to your colleagues.
Jessica
They can access the full report even if they aren’t registered pCloudy users.
Jessica
That concludes this brief look at the failure analysis capabilities.
Jessica
That also brings us to the end of the live demonstration.
Jessica
I’ve shown how a real workflow can be managed through Claude, including retrieving results and reports.
Jessica
You can also ask Claude to analyze failures and view the entire workflow in one conversation.
Jessica
That workflow extends from booking a device and creating test cases through to execution, results, and analysis.
Jessica
Now, let’s look at how to get started.
Jessica
The setup is straightforward and takes only a few minutes.
Jessica
First, you need a pCloudy account and an AI tool such as Claude or Cursor.
Jessica
You then configure the Qpilot MCP server, connect it to your chosen tool, and begin generating tests and analyzing results.
Jessica
We provide complete documentation with all the required prerequisites and setup instructions.
Jessica
We’ll be happy to share it so you can connect MCP to your pCloudy account whenever you’re ready.
Jessica
Here, you can see the documentation for configuring the MCP server with pCloudy.
Jessica
It includes the prerequisites and exact steps required to complete the setup.
Jessica
I hope this gives you a clear understanding of what you can achieve and how you can optimize your workflow by connecting MCP with Qpilot and pCloudy.
Jessica
Please feel free to contact us if you have any questions or would like another demonstration or clarification.
Jessica
With that, we can move to the Q&A session.
Jessica
I don’t see any questions yet, so please add them using the Q&A tab.
Jessica
I can see that our first question has arrived.
Jessica
Do I need coding knowledge to use Claude with Qpilot MCP, or can a manual tester use it?
Jessica
Manual testers can definitely use it.
Jessica
Qpilot itself doesn’t require any coding knowledge.
Jessica
Once you follow the setup instructions and establish the MCP connection, you don’t need coding skills to use it.
Jessica
You can perform both automated and manual testing without writing code.
Jessica
How long does the setup take?
Jessica
It doesn’t take long; the setup requires only a few minutes.
Jessica
In less than half an hour, you can be up and running and create your first test script through Claude and MCP.
Jessica
Does this work with our existing test scripts, or do we have to start from scratch?
Jessica
You don’t need to start from scratch.
Dinakar
Without further ado, I’ll hand it over to Jessica.
Dinakar
Over to you, Jessica.
Jessica
Thank you so much for that introduction, Dinakar.
Jessica
I hope my screen is visible.
Jessica
As Dinakar mentioned, today’s topic is how Claude and the Qpilot MCP connection can help you generate tests faster, view all your test results in one place, and save hours otherwise spent switching between tools.
Jessica
Let’s begin by looking at what AI has already solved and why some testing activities still take so much time.
Jessica
Test generation, for example, has largely been solved by recent advances in AI.
Jessica
Whether you use an AI agent such as Qpilot or GitHub Copilot to write Playwright scripts, you can describe a flow and generate scripts quickly.
Jessica
That’s genuinely impressive, and the capability has become widely available very quickly.
Jessica
However, most QA teams still use several different tools.
Jessica
A QA engineer might use one tool to write a test, another to share the result, an AI assistant, an editor or IDE, a device farm, and separate dashboards and reports.
Jessica
They may then log bugs or tasks in Slack or Jira.
Jessica
This creates a significant amount of context switching.
Jessica
If API testing is involved, the engineer may also use a tool such as Postman.
Jessica
Debugging a single failure can require at least three or four manual handoffs.
Jessica
Each action and result must also be recorded separately in the relevant tool.
Jessica
This is where teams lose hours during every sprint or release cycle.
Jessica
As people switch between tools, important details are often lost, so their AI tools never receive the full context.
Jessica
Teams also spend time manually reviewing logs, dashboards, and reports.
Jessica
Meanwhile, much of their innovative work is happening inside AI tools such as Cursor or Claude Code.
Jessica
They can’t easily bring the context from other tools back into the environment they primarily use to develop scripts and generate new builds.
Jessica
That is the central problem the Qpilot and Claude MCP connection is designed to solve.
Jessica
Let me summarize the key agenda for today’s session.
Jessica
First, you’ll learn how Claude connects with Qpilot through MCP.
Jessica
You’ll see how to access pCloudy’s real devices directly from tools such as Cursor or Claude.
Jessica
Second, we’ll walk through a live demonstration.
Jessica
Third, I’ll cover our AI-powered test failure analysis reports and show how they significantly reduce debugging time.
Jessica
You’ll see how the system analyzes logs and provides detailed findings quickly.
Jessica
I’ll also explain how to execute test cases and create a reusable MCP setup for your team.
Jessica
As mentioned earlier, we’ll answer questions at the end of the demonstration.
Jessica
Let’s begin with the first section: where QA teams are still losing time and context between tools.
Jessica
In a typical workflow, you often begin with an AI assistant such as Cursor, GitHub Copilot, or Claude Code.
Jessica
You describe the flow you want to test, refine the output through a few exchanges, and eventually receive a script that is ready to run.
Jessica
Next, you copy the script from the chat window and paste it into your codebase.
Jessica
You then fix imports, apply page-object conventions, and replace locators where necessary.
Jessica
This can create further back-and-forth work in your IDE or test editor.
Jessica
When it’s time to use a device farm, you must book a device, add the required capabilities, connect to the device, and run your test cases.
Jessica
You may switch between your IDE or CI/CD tool and a device farm such as pCloudy to track execution.
Jessica
You may also open a separate dashboarding solution to review the results.
Jessica
Finally, you update documentation or log bugs in tools such as Slack or Jira.
Jessica
In this workflow, you’ve switched between five tools and created five opportunities to lose context.
Jessica
You may lose critical details or fixes made in the test editor and fail to bring that context back into your AI assistant.
Jessica
This limits your ability to improve the accuracy of the scripts it generates.
Jessica
That lost context eventually translates into a significant loss of time.
Jessica
Context switching is only one part of the problem.
Jessica
Teams also spend time booking devices, uploading builds, and tracking execution status.
Jessica
Each time, you may need to open a dashboard or active session to check whether execution has finished and whether the report has been generated.
Jessica
These may seem like simple tasks, but they accumulate and consume a meaningful amount of time.
Jessica
Another time-consuming task is reviewing raw logs.
Jessica
When a test case fails, you may need to examine extensive logs to understand what went wrong.
Jessica
With pCloudy’s AI failure analysis agent, you can immediately see the relevant log lines and identify where the test failed.
Jessica
Reporting can create another delay, especially when test results are stored in a separate tool.
Jessica
You still need to retrieve the report, debug the issue, and send the findings back to Jira or Slack.
Jessica
These are the main touchpoints where teams lose both time and context.
Jessica
Now, let me introduce the Claude and Qpilot MCP connection.
Jessica
Claude is the focus of today’s demonstration, but pCloudy also integrates with tools such as Cursor.
Jessica
MCP stands for Model Context Protocol.
Jessica
It attracted considerable attention a few months ago and is an open standard introduced by Anthropic.
Jessica
Put simply, MCP allows an AI assistant to communicate directly with an external system instead of requiring you to act as the intermediary.
Jessica
It enables the assistant to retrieve data from your tools and perform actions on your platform.
Jessica
That is the core idea behind an MCP connection.
Jessica
It provides a common protocol for interacting with different tools and simplifies those connections.
Jessica
You can think of it as an API-like connection between Claude and pCloudy.
Jessica
Instead of reading information in one system, retyping it into another, and carrying the context in your head, MCP connects the systems directly.
Jessica
It removes you from the relay loop and lets your AI assistant communicate directly with the platform.
Jessica
If you haven’t heard of Qpilot, it is pCloudy’s AI-based test creation solution.
Jessica
You can explore our other Qpilot webinars, watch demonstrations on our website, or contact us to learn more.
Jessica
Qpilot is an AI-powered test creation platform built on top of pCloudy’s real-device infrastructure.
Jessica
pCloudy provides real Android and iOS devices, as well as a browser cloud, for manual and automated mobile and web testing.
Jessica
You can run tests using Appium, Selenium, Playwright, and other popular frameworks.
Jessica
Alternatively, you can use Qpilot, our AI layer, to convert plain-English descriptions into automated test steps.
Jessica
Real-device infrastructure is the foundation of the pCloudy platform.
Jessica
We host physical devices, not simulators or emulators, in our device cloud.
Jessica
Through Claude, you can access both Qpilot’s AI capabilities and pCloudy’s real-device infrastructure.
Jessica
Let’s look at some of the practical tasks this connection can support.
Jessica
First, you can generate tests directly from Claude instead of opening Qpilot separately.
Jessica
Once the pCloudy MCP connection is enabled, you can describe the test you need, ask Claude to create it, or retrieve an existing test case.
Jessica
You can also retrieve results and reports, and monitor live executions.
Jessica
This applies to both automated and manual sessions.
Jessica
Because Claude can access those reports, it can analyze failures using their complete context.
Jessica
You don’t need to copy and paste log lines or manually provide additional background information.
Jessica
Claude can retrieve the required context directly from the reports.
Jessica
You can also orchestrate and run collections of multiple test cases directly through the connection.
Jessica
Now, let’s compare the workflow with and without MCP.
Jessica
Without MCP, one task may require five separate tools—one for test generation, another for results, another for analysis and reporting, and another for device booking.
Jessica
Each activity is isolated in its own silo.
Jessica
Manually booking devices and tracking execution only addresses part of the problem because test execution is still triggered separately from Qpilot.
Jessica
MCP changes this by bringing the end-to-end workflow into the AI tool you’re already using.
Jessica
It also eliminates the need to debug failures one log line at a time.
Jessica
pCloudy’s AI failure analysis identifies the relevant issues for you.
Jessica
Claude can then provide a shareable report link and use the logs and failure details as context.
Jessica
These are some of the key advantages of using MCP to create and run tests, book devices, and manage manual or automated sessions.
Jessica
Even simple platform queries can otherwise take up time.
Jessica
For example, you can ask Claude to list available Android 15 devices, Google devices running Android 14 or 16, or any other configuration you need.
Jessica
It can return the available devices and the applications associated with your account.
Jessica
You can also ask for the latest build uploaded to pCloudy.
Jessica
This means you don’t need to log in every time you need basic information about the platform or its real devices.
Jessica
That saves time and helps you begin testing faster.
Jessica
Next, let’s look at our AI-powered failure analysis reports.
Jessica
These reports automatically transform raw logs into root-cause insights at both the test-case and build levels.
Jessica
They provide context for failures such as missing elements and performance problems.
Jessica
If a test experiences lag or latency, the report can identify possible causes.
Jessica
It also provides recommendations and suggested fixes.
Jessica
This AI-powered report is generated automatically for every test run.
Jessica
You receive analysis at both the build level and the individual test-case level.
Jessica
Let’s compare manual debugging with debugging through Claude and Qpilot MCP.
Jessica
In a manual workflow, you open the dashboard, locate a failed run, and cross-reference logs against screenshots or video.
Jessica
You then determine which video frame corresponds to each log line and develop a theory about the root cause.
Jessica
This can take a significant amount of time.
Jessica
You may also need to collaborate with multiple teams without a single source of truth.
Jessica
Debugging one issue can consume an entire afternoon or even a full day.
Jessica
With Claude and Qpilot MCP, Claude retrieves the failed-run logs and screenshots for you.
Jessica
Because pCloudy’s reporting already includes AI-generated root-cause analysis and the most relevant log lines, Claude can provide the context you need to investigate further.
Jessica
What matters is the relationship between the different pieces of evidence, rather than each source operating in isolation.
Jessica
Correlating that data manually is tedious and requires considerable switching between tools.
Jessica
This is where the connected workflow delivers a meaningful benefit.
Jessica
You can also generate a shareable summary and refer back to it later.
Jessica
If a test fails again, you can connect it with an earlier error and use the complete history to debug the issue more efficiently.
Jessica
With that, let’s move into the live demonstration of the Claude MCP solution.
Jessica
Before I demonstrate the MCP connection, I’ll briefly introduce pCloudy for anyone who isn’t familiar with our platform.
Jessica
pCloudy is a device cloud with intelligence and a SaaS-based platform that can be accessed from anywhere.
Jessica
At the core of our solution, we host real mobile devices and browsers.
Jessica
We operate data centers across multiple regions, allowing you to connect to devices, launch applications or URLs, and begin testing.
Jessica
In other words, we provide the test infrastructure.
Jessica
This is the landing page of the pCloudy platform.
Jessica
Every device listed here is a real, physical device hosted in one of our data-center locations.
Jessica
You can see the available locations, upload your builds, and start testing.
Jessica
The same process applies to our iOS devices.
Jessica
We also provide browser infrastructure with Windows and macOS machines.
Jessica
You can connect to different browsers and versions for cross-browser testing.
Jessica
The platform also supports integrations with popular automation frameworks.
Jessica
These include Appium, Selenium, Playwright, and several other tools listed in our documentation.
Jessica
Now, let’s look at Qpilot.
Jessica
Qpilot allows you to create automated test cases using plain-English descriptions.
Jessica
You can perform all these actions through the Qpilot interface.
Jessica
However, today I’ll focus on how you can create these tests without logging in to pCloudy.
Jessica
Qpilot is an AI-based test creation agent.
Jessica
You don’t need to learn special keywords; you simply describe the test in plain English.
Jessica
Qpilot converts that description into an end-to-end test case.
Jessica
It supports web, iOS, Android, and API testing on a single platform.
Jessica
Qpilot itself consolidates several tools.
Jessica
You might otherwise use one tool for API testing and another as a device farm.
Jessica
Qpilot combines pCloudy’s real-device infrastructure, an AI test creation agent, reporting, API testing, and integrations with Jira, Trello, and Slack.
Jessica
This already delivers significant tool consolidation, and MCP extends those benefits further.
Jessica
Now, let’s move to the Claude MCP demonstration.
Jessica
Here is my Claude account, and we provide documentation explaining how to enable MCP connections.
Jessica
In my list of connectors, you can see that the pCloudy toggle is already enabled.
Jessica
This means there is an active connection, and Claude can retrieve details from pCloudy and perform supported actions.
Jessica
The connection covers both pCloudy’s real-device infrastructure and Qpilot’s test creation capabilities.
Jessica
Let me begin with a simple request.
Jessica
I’ll ask Claude to list the available Google devices running Android versions 15 and 16.
Jessica
I’ve entered that request in plain English.
Jessica
Claude is now identifying the appropriate tools and retrieving the device list.
Jessica
These requests are being sent to pCloudy so Claude can obtain the relevant context.
Jessica
You can now see the available devices listed in the response.
Jessica
Similarly, I can ask Claude to list the available iOS devices.
Jessica
It retrieves the available configurations and operating-system versions for me.
Jessica
If you want to know whether a particular device is available or busy, or which devices can be used for automation, you can ask directly in Claude.
Jessica
You can even book a device through the same conversation.
Jessica
For example, I’ll ask Claude to connect to a Samsung device running Android 14 and install the IKEA app.
Jessica
Once I submit the instruction, Claude connects to the platform and begins performing the requested actions.
Jessica
A small amount of processing time is expected while Claude communicates with pCloudy.
Jessica
You can see each API call and action taking place behind the scenes while the device is being booked.
Jessica
Several individual actions are completed as part of this single request.
Jessica
I’ll return to pCloudy and check the active sessions.
Jessica
You can see that the platform has already selected the device and the connection is in progress.
Jessica
This should give you an idea of the interactions you can perform once the MCP connection is running.
Jessica
I’d also like to show how the connection works with the Qpilot agent.
Jessica
Here, I’ve asked Claude to create a new test case.
Jessica
I provided the test-case name, the APK or IPA name, and all the other information required to create it.
Jessica
I also specified that the test should be generated on a Google Pixel 2 XL.
Jessica
Claude is processing the request and completing the required action at each step.
Jessica
It first found the demo project, then located the appropriate folder and retrieved its folder ID.
Jessica
Once it has everything it needs, it creates the test case within that folder.
Jessica
The test case has now been created, and Claude has returned the session ID and a summary of the completed actions.
Jessica
Claude also suggests the next action and asks whether it should add more steps to the test case.
Jessica
I’ve responded with another plain-English description.
Jessica
You can see all seven steps being added to the test case.
Jessica
Claude has also summarized the outcome and confirmed that the test ran end to end on the Pixel 2 XL without any issues.
Jessica
You can also edit an existing test case.
Jessica
If you need to add a few steps, simply ask Claude to connect to the device and update the specified test case.
Jessica
Claude opens the test, adds the requested steps, and completes the update.
Jessica
This makes it easy to maintain and update existing test cases.
Jessica
On the Qpilot side, you can now see the test case with the exact name I specified in Claude.
Jessica
The automated test script has also been created.
Jessica
I can run it from either Claude or Qpilot.
Jessica
I can execute the test cases and retrieve the results through the same connected workflow.
Jessica
The test is now running, and I can monitor it live.
Jessica
Claude can provide links to the live execution and the final reports.
Jessica
Now, let’s look at the second part of the demonstration: rapid debugging with our AI failure analysis agent.
Jessica
Every automated test you run includes detailed reporting within the pCloudy platform.
Jessica
Here, you can see the most recent execution.
Jessica
When I open the build, I can view the status of every test case, including how many have completed, failed, timed out, or remain in progress.
Jessica
The platform also provides build-level analysis.
Jessica
It shows which sessions encountered an element-not-found error, which experienced slow page loads, and which devices had issues.
Jessica
Across past runs, you can identify devices with a higher tendency to fail and understand the errors associated with them.
Jessica
This allows you to debug at the build or project level.
Jessica
You can also open an individual test case for more detailed analysis.
Jessica
The text logs show the output of each step, including the clicks and actions performed.
Jessica
The step view shows every test step, the exact point of failure, and the corresponding error.
Jessica
If you need more detail, pCloudy still provides access to the complete raw and device logs.
Jessica
However, the AI analysis fast-tracks the investigation.
Jessica
In this example, the test failed because an element could not be found.
Jessica
The failure matches the error shown in the step view, and the analysis highlights the relevant log line.
Jessica
It identifies the exact log entry where the error occurred and recommends an action.
Jessica
In this case, the recommendation is to validate the element’s XPath.
Jessica
The report helps us pinpoint that the XPath used to select a location from the drop-down menu was probably incorrect.
Jessica
We can therefore identify the likely root cause without spending time searching through the logs manually.
Jessica
pCloudy also provides access to the complete video and screenshots.
Jessica
You can log bugs directly from the platform using integrations with Jira, Trello, Slack, and GitHub.
Jessica
The reports are shareable, which makes it easier to provide Claude with the relevant context.
Jessica
You can generate a shareable link and send it to your colleagues.
Jessica
They can access the full report even if they aren’t registered pCloudy users.
Jessica
That concludes this brief look at the failure analysis capabilities.
Jessica
That also brings us to the end of the live demonstration.
Jessica
I’ve shown how a real workflow can be managed through Claude, including retrieving results and reports.
Jessica
You can also ask Claude to analyze failures and view the entire workflow in one conversation.
Jessica
That workflow extends from booking a device and creating test cases through to execution, results, and analysis.
Jessica
Now, let’s look at how to get started.
Jessica
The setup is straightforward and takes only a few minutes.
Jessica
First, you need a pCloudy account and an AI tool such as Claude or Cursor.
Jessica
You then configure the Qpilot MCP server, connect it to your chosen tool, and begin generating tests and analyzing results.
Jessica
We provide complete documentation with all the required prerequisites and setup instructions.
Jessica
We’ll be happy to share it so you can connect MCP to your pCloudy account whenever you’re ready.
Jessica
Here, you can see the documentation for configuring the MCP server with pCloudy.
Jessica
It includes the prerequisites and exact steps required to complete the setup.
Jessica
I hope this gives you a clear understanding of what you can achieve and how you can optimize your workflow by connecting MCP with Qpilot and pCloudy.
Jessica
Please feel free to contact us if you have any questions or would like another demonstration or clarification.
Jessica
With that, we can move to the Q&A session.
Jessica
I don’t see any questions yet, so please add them using the Q&A tab.
Jessica
I can see that our first question has arrived.
Jessica
Do I need coding knowledge to use Claude with Qpilot MCP, or can a manual tester use it?
Jessica
Manual testers can definitely use it.
Jessica
Qpilot itself doesn’t require any coding knowledge.
Jessica
Once you follow the setup instructions and establish the MCP connection, you don’t need coding skills to use it.
Jessica
You can perform both automated and manual testing without writing code.
Jessica
How long does the setup take?
Jessica
It doesn’t take long; the setup requires only a few minutes.
Jessica
In less than half an hour, you can be up and running and create your first test script through Claude and MCP.
Jessica
Does this work with our existing test scripts, or do we have to start from scratch?
Jessica
You don’t need to start from scratch.
Jessica
In a typical workflow, you often begin with an AI assistant such as Cursor, GitHub Copilot, or Claude Code.
Jessica
You describe the flow you want to test, refine the output through a few exchanges, and eventually receive a script that is ready to run.
Jessica
Next, you copy the script from the chat window and paste it into your codebase.
Jessica
You then fix imports, apply page-object conventions, and replace locators where necessary.
Jessica
This can create further back-and-forth work in your IDE or test editor.
Jessica
When it’s time to use a device farm, you must book a device, add the required capabilities, connect to the device, and run your test cases.
Jessica
You may switch between your IDE or CI/CD tool and a device farm such as pCloudy to track execution.
Jessica
You may also open a separate dashboarding solution to review the results.
Jessica
Finally, you update documentation or log bugs in tools such as Slack or Jira.
Jessica
In this workflow, you’ve switched between five tools and created five opportunities to lose context.
Jessica
You may lose critical details or fixes made in the test editor and fail to bring that context back into your AI assistant.
Jessica
This limits your ability to improve the accuracy of the scripts it generates.
Jessica
That lost context eventually translates into a significant loss of time.
Jessica
Context switching is only one part of the problem.
Jessica
Teams also spend time booking devices, uploading builds, and tracking execution status.
Jessica
Each time, you may need to open a dashboard or active session to check whether execution has finished and whether the report has been generated.
Jessica
These may seem like simple tasks, but they accumulate and consume a meaningful amount of time.
Jessica
Another time-consuming task is reviewing raw logs.
Jessica
When a test case fails, you may need to examine extensive logs to understand what went wrong.
Jessica
With pCloudy’s AI failure analysis agent, you can immediately see the relevant log lines and identify where the test failed.
Jessica
Reporting can create another delay, especially when test results are stored in a separate tool.
Jessica
You still need to retrieve the report, debug the issue, and send the findings back to Jira or Slack.
Jessica
These are the main touchpoints where teams lose both time and context.
Jessica
Now, let me introduce the Claude and Qpilot MCP connection.
Jessica
Claude is the focus of today’s demonstration, but pCloudy also integrates with tools such as Cursor.
Jessica
MCP stands for Model Context Protocol.
Jessica
It attracted considerable attention a few months ago and is an open standard introduced by Anthropic.
Jessica
Put simply, MCP allows an AI assistant to communicate directly with an external system instead of requiring you to act as the intermediary.
Jessica
It enables the assistant to retrieve data from your tools and perform actions on your platform.
Jessica
That is the core idea behind an MCP connection.
Jessica
It provides a common protocol for interacting with different tools and simplifies those connections.
Jessica
You can think of it as an API-like connection between Claude and pCloudy.
Jessica
Instead of reading information in one system, retyping it into another, and carrying the context in your head, MCP connects the systems directly.
Jessica
It removes you from the relay loop and lets your AI assistant communicate directly with the platform.
Jessica
If you haven’t heard of Qpilot, it is pCloudy’s AI-based test creation solution.
Jessica
You can explore our other Qpilot webinars, watch demonstrations on our website, or contact us to learn more.
Jessica
Qpilot is an AI-powered test creation platform built on top of pCloudy’s real-device infrastructure.
Jessica
pCloudy provides real Android and iOS devices, as well as a browser cloud, for manual and automated mobile and web testing.
Jessica
You can run tests using Appium, Selenium, Playwright, and other popular frameworks.
Jessica
Alternatively, you can use Qpilot, our AI layer, to convert plain-English descriptions into automated test steps.
Jessica
Real-device infrastructure is the foundation of the pCloudy platform.
Jessica
We host physical devices, not simulators or emulators, in our device cloud.
Jessica
Through Claude, you can access both Qpilot’s AI capabilities and pCloudy’s real-device infrastructure.
Jessica
Let’s look at some of the practical tasks this connection can support.
Jessica
First, you can generate tests directly from Claude instead of opening Qpilot separately.
Jessica
Once the pCloudy MCP connection is enabled, you can describe the test you need, ask Claude to create it, or retrieve an existing test case.
Jessica
You can also retrieve results and reports, and monitor live executions.
Jessica
This applies to both automated and manual sessions.
Jessica
Because Claude can access those reports, it can analyze failures using their complete context.
Jessica
You don’t need to copy and paste log lines or manually provide additional background information.
Jessica
Claude can retrieve the required context directly from the reports.
Jessica
You can also orchestrate and run collections of multiple test cases directly through the connection.
Jessica
Now, let’s compare the workflow with and without MCP.
Jessica
Without MCP, one task may require five separate tools—one for test generation, another for results, another for analysis and reporting, and another for device booking.
Jessica
Each activity is isolated in its own silo.
Jessica
Manually booking devices and tracking execution only addresses part of the problem because test execution is still triggered separately from Qpilot.
Jessica
MCP changes this by bringing the end-to-end workflow into the AI tool you’re already using.
Jessica
It also eliminates the need to debug failures one log line at a time.
Jessica
pCloudy’s AI failure analysis identifies the relevant issues for you.
Jessica
Claude can then provide a shareable report link and use the logs and failure details as context.
Jessica
These are some of the key advantages of using MCP to create and run tests, book devices, and manage manual or automated sessions.
Jessica
Even simple platform queries can otherwise take up time.
Jessica
For example, you can ask Claude to list available Android 15 devices, Google devices running Android 14 or 16, or any other configuration you need.
Jessica
It can return the available devices and the applications associated with your account.
Jessica
You can also ask for the latest build uploaded to pCloudy.
Jessica
This means you don’t need to log in every time you need basic information about the platform or its real devices.
Jessica
That saves time and helps you begin testing faster.
Jessica
Next, let’s look at our AI-powered failure analysis reports.
Jessica
These reports automatically transform raw logs into root-cause insights at both the test-case and build levels.
Jessica
They provide context for failures such as missing elements and performance problems.
Jessica
If a test experiences lag or latency, the report can identify possible causes.
Jessica
It also provides recommendations and suggested fixes.
Jessica
This AI-powered report is generated automatically for every test run.
Jessica
You receive analysis at both the build level and the individual test-case level.
Jessica
Let’s compare manual debugging with debugging through Claude and Qpilot MCP.
Jessica
In a manual workflow, you open the dashboard, locate a failed run, and cross-reference logs against screenshots or video.
Jessica
You then determine which video frame corresponds to each log line and develop a theory about the root cause.
Jessica
This can take a significant amount of time.
Jessica
You may also need to collaborate with multiple teams without a single source of truth.
Jessica
Debugging one issue can consume an entire afternoon or even a full day.
Jessica
With Claude and Qpilot MCP, Claude retrieves the failed-run logs and screenshots for you.
Jessica
Because pCloudy’s reporting already includes AI-generated root-cause analysis and the most relevant log lines, Claude can provide the context you need to investigate further.
Jessica
What matters is the relationship between the different pieces of evidence, rather than each source operating in isolation.
Jessica
Correlating that data manually is tedious and requires considerable switching between tools.
Jessica
This is where the connected workflow delivers a meaningful benefit.
Jessica
You can also generate a shareable summary and refer back to it later.
Jessica
If a test fails again, you can connect it with an earlier error and use the complete history to debug the issue more efficiently.
Jessica
With that, let’s move into the live demonstration of the Claude MCP solution.
Jessica
Before I demonstrate the MCP connection, I’ll briefly introduce pCloudy for anyone who isn’t familiar with our platform.
Jessica
pCloudy is a device cloud with intelligence and a SaaS-based platform that can be accessed from anywhere.
Jessica
At the core of our solution, we host real mobile devices and browsers.
Jessica
We operate data centers across multiple regions, allowing you to connect to devices, launch applications or URLs, and begin testing.
Jessica
In other words, we provide the test infrastructure.
Jessica
This is the landing page of the pCloudy platform.
Jessica
Every device listed here is a real, physical device hosted in one of our data-center locations.
Jessica
You can see the available locations, upload your builds, and start testing.
Jessica
The same process applies to our iOS devices.
Jessica
We also provide browser infrastructure with Windows and macOS machines.
Jessica
You can connect to different browsers and versions for cross-browser testing.
Jessica
The platform also supports integrations with popular automation frameworks.
Jessica
These include Appium, Selenium, Playwright, and several other tools listed in our documentation.
Jessica
Now, let’s look at Qpilot.
Jessica
Qpilot allows you to create automated test cases using plain-English descriptions.
Jessica
You can perform all these actions through the Qpilot interface.
Jessica
However, today I’ll focus on how you can create these tests without logging in to pCloudy.
Jessica
Qpilot is an AI-based test creation agent.
Jessica
You don’t need to learn special keywords; you simply describe the test in plain English.
Jessica
Qpilot converts that description into an end-to-end test case.
Jessica
It supports web, iOS, Android, and API testing on a single platform.
Jessica
Qpilot itself consolidates several tools.
Jessica
You might otherwise use one tool for API testing and another as a device farm.
Jessica
Qpilot combines pCloudy’s real-device infrastructure, an AI test creation agent, reporting, API testing, and integrations with Jira, Trello, and Slack.
Jessica
This already delivers significant tool consolidation, and MCP extends those benefits further.
Jessica
Now, let’s move to the Claude MCP demonstration.
Jessica
Here is my Claude account, and we provide documentation explaining how to enable MCP connections.
Jessica
In my list of connectors, you can see that the pCloudy toggle is already enabled.
Jessica
This means there is an active connection, and Claude can retrieve details from pCloudy and perform supported actions.
Jessica
The connection covers both pCloudy’s real-device infrastructure and Qpilot’s test creation capabilities.
Jessica
Let me begin with a simple request.
Jessica
I’ll ask Claude to list the available Google devices running Android versions 15 and 16.
Jessica
I’ve entered that request in plain English.
Jessica
Claude is now identifying the appropriate tools and retrieving the device list.
Jessica
These requests are being sent to pCloudy so Claude can obtain the relevant context.
Jessica
You can now see the available devices listed in the response.
Jessica
Similarly, I can ask Claude to list the available iOS devices.
Jessica
It retrieves the available configurations and operating-system versions for me.
Jessica
If you want to know whether a particular device is available or busy, or which devices can be used for automation, you can ask directly in Claude.
Jessica
You can even book a device through the same conversation.
Jessica
For example, I’ll ask Claude to connect to a Samsung device running Android 14 and install the IKEA app.
Jessica
Once I submit the instruction, Claude connects to the platform and begins performing the requested actions.
Jessica
A small amount of processing time is expected while Claude communicates with pCloudy.
Jessica
You can see each API call and action taking place behind the scenes while the device is being booked.
Jessica
Several individual actions are completed as part of this single request.
Jessica
I’ll return to pCloudy and check the active sessions.
Jessica
You can see that the platform has already selected the device and the connection is in progress.
Jessica
This should give you an idea of the interactions you can perform once the MCP connection is running.
Jessica
I’d also like to show how the connection works with the Qpilot agent.
Jessica
Here, I’ve asked Claude to create a new test case.
Jessica
I provided the test-case name, the APK or IPA name, and all the other information required to create it.
Jessica
I also specified that the test should be generated on a Google Pixel 2 XL.
Jessica
Claude is processing the request and completing the required action at each step.
Jessica
It first found the demo project, then located the appropriate folder and retrieved its folder ID.
Jessica
Once it has everything it needs, it creates the test case within that folder.
Jessica
The test case has now been created, and Claude has returned the session ID and a summary of the completed actions.
Jessica
Claude also suggests the next action and asks whether it should add more steps to the test case.
Jessica
I’ve responded with another plain-English description.
Jessica
You can see all seven steps being added to the test case.
Jessica
Claude has also summarized the outcome and confirmed that the test ran end to end on the Pixel 2 XL without any issues.
Jessica
You can also edit an existing test case.
Jessica
If you need to add a few steps, simply ask Claude to connect to the device and update the specified test case.
Jessica
Claude opens the test, adds the requested steps, and completes the update.
Jessica
This makes it easy to maintain and update existing test cases.
Jessica
On the Qpilot side, you can now see the test case with the exact name I specified in Claude.
Jessica
The automated test script has also been created.
Jessica
I can run it from either Claude or Qpilot.
Jessica
I can execute the test cases and retrieve the results through the same connected workflow.
Jessica
The test is now running, and I can monitor it live.
Jessica
Claude can provide links to the live execution and the final reports.
Jessica
Now, let’s look at the second part of the demonstration: rapid debugging with our AI failure analysis agent.
Jessica
Every automated test you run includes detailed reporting within the pCloudy platform.
Jessica
Here, you can see the most recent execution.
Jessica
When I open the build, I can view the status of every test case, including how many have completed, failed, timed out, or remain in progress.
Jessica
The platform also provides build-level analysis.
Jessica
It shows which sessions encountered an element-not-found error, which experienced slow page loads, and which devices had issues.
Jessica
Across past runs, you can identify devices with a higher tendency to fail and understand the errors associated with them.
Jessica
This allows you to debug at the build or project level.
Jessica
You can also open an individual test case for more detailed analysis.
Jessica
The text logs show the output of each step, including the clicks and actions performed.
Jessica
The step view shows every test step, the exact point of failure, and the corresponding error.
Jessica
If you need more detail, pCloudy still provides access to the complete raw and device logs.
Jessica
However, the AI analysis fast-tracks the investigation.
Jessica
In this example, the test failed because an element could not be found.
Jessica
The failure matches the error shown in the step view, and the analysis highlights the relevant log line.
Jessica
It identifies the exact log entry where the error occurred and recommends an action.
Jessica
In this case, the recommendation is to validate the element’s XPath.
Jessica
The report helps us pinpoint that the XPath used to select a location from the drop-down menu was probably incorrect.
Jessica
We can therefore identify the likely root cause without spending time searching through the logs manually.
Jessica
pCloudy also provides access to the complete video and screenshots.
Jessica
You can log bugs directly from the platform using integrations with Jira, Trello, Slack, and GitHub.
Jessica
The reports are shareable, which makes it easier to provide Claude with the relevant context.
Jessica
You can generate a shareable link and send it to your colleagues.
Jessica
They can access the full report even if they aren’t registered pCloudy users.
Jessica
That concludes this brief look at the failure analysis capabilities.
Jessica
That also brings us to the end of the live demonstration.
Jessica
I’ve shown how a real workflow can be managed through Claude, including retrieving results and reports.
Jessica
You can also ask Claude to analyze failures and view the entire workflow in one conversation.
Jessica
That workflow extends from booking a device and creating test cases through to execution, results, and analysis.
Jessica
Now, let’s look at how to get started.
Jessica
The setup is straightforward and takes only a few minutes.
Jessica
First, you need a pCloudy account and an AI tool such as Claude or Cursor.
Jessica
You then configure the Qpilot MCP server, connect it to your chosen tool, and begin generating tests and analyzing results.
Jessica
We provide complete documentation with all the required prerequisites and setup instructions.
Jessica
We’ll be happy to share it so you can connect MCP to your pCloudy account whenever you’re ready.
Jessica
Here, you can see the documentation for configuring the MCP server with pCloudy.
Jessica
It includes the prerequisites and exact steps required to complete the setup.
Jessica
I hope this gives you a clear understanding of what you can achieve and how you can optimize your workflow by connecting MCP with Qpilot and pCloudy.
Jessica
Please feel free to contact us if you have any questions or would like another demonstration or clarification.
Jessica
With that, we can move to the Q&A session.
Jessica
I don’t see any questions yet, so please add them using the Q&A tab.
Jessica
I can see that our first question has arrived.
Jessica
Do I need coding knowledge to use Claude with Qpilot MCP, or can a manual tester use it?
Jessica
Manual testers can definitely use it.
Jessica
Qpilot itself doesn’t require any coding knowledge.
Jessica
Once you follow the setup instructions and establish the MCP connection, you don’t need coding skills to use it.
Jessica
You can perform both automated and manual testing without writing code.
Jessica
How long does the setup take?
Jessica
It doesn’t take long; the setup requires only a few minutes.
Jessica
In less than half an hour, you can be up and running and create your first test script through Claude and MCP.
Jessica
Does this work with our existing test scripts, or do we have to start from scratch?
Jessica
You don’t need to start from scratch.
Dinakar
If you have any further questions, you can contact us through the chat option on our website or email us at info@pcloudy.com.
Dinakar
Please feel free to reach out, and we’ll respond as soon as possible.
Dinakar
Thank you once again for being such an engaged audience, and happy testing.
