Dinakar
All right.
Dinakar
Hello everyone and welcome to Pcloudy's webinar on unlock autonomous testing with agentic AI Qpilot.ai.
Dinakar
My name is Dinakar and I'll be the host for today.
Dinakar
Before we get started, I would like to go over a few housekeeping rules.
Dinakar
You may have noticed that your line is currently muted.
Dinakar
However, you can submit your questions during the webinar using the Q&A option available.
Dinakar
We will answer your questions as part of the Q&A section at the end of the webinar.
Dinakar
Also, do note that this webinar is being recorded and we will send out the link of the recording to watch again or share with your colleagues later.
Dinakar
With that said, let me introduce the speaker for today.
Dinakar
He needs no introduction, Avinash Tiwari.
Dinakar
Avinash is a renowned thought leader, a recognized keynote speaker, and the co-founder of Pcloudy.
Dinakar
He has around 20 plus years of experience in product development and testing.
Dinakar
And he brings with him a passion for emerging technology and quick adoption.
Dinakar
Both of which which have solidified his reputation as a thought leader in the app testing space.
Dinakar
So, without further ado, let me hand it to our speaker, Avinash Tiwari.
Avinash
Thanks, Dinakar.
Avinash
Thanks for the introduction.
Avinash
And a warm welcome to all of you again on behalf of myself and my colleague Shoaib.
Avinash
We are very very excited to be here and present to you what we have created uh uh as as Qpilot.ai.
Avinash
Um so, over next 45 minutes myself and Shoaib will be taking you through a ride of a kind.
Avinash
So, please fasten your seat belts and we are ready to take take off.
Avinash
Um So initially I'll talk about what this beast is all about.
Avinash
And then Shoaib will take it over and and start showing you how it works.
Avinash
Uh so the beast in action.
Avinash
Um that's the approach we have.
Avinash
So please be settled and and we are all set to go.
Avinash
Before I jump in I want to take a step back and talk about the buzz around AI, GenAI, or agentic AI.
Avinash
So a few years back uh I think we were all talking about transformation of organizations to become a digital first tech company.
Avinash
And we saw companies like Uber, Tesla emerging in this world.
Avinash
Uh so do we know Uber as a taxi company?
Avinash
I guess we all know Uber as an app company.
Avinash
Same with Tesla.
Avinash
Do we know it as a car company?
Avinash
We all know it as a tech powerhouse uh company.
Avinash
And similarly we saw in different industries banks becoming tech first companies.
Avinash
And that uh gave rise to the evolution of developer and test tools.
Avinash
Uh because with so much tech inside every organization uh you can't do anything without right set of developer and test tools.
Avinash
And we'll talk about that as we go further.
Avinash
I was recently uh hearing a talk from Satya Nadella and he talked about the current situation.
Avinash
What what we are talking about today.
Avinash
Every company has to become an AI company.
Avinash
And we are already seeing that in action.
Avinash
And testing has to follow the trend.
Avinash
Like we followed the trend when the companies became the digital first uh tech company.
Avinash
So with that as as background, we have seen the evolution of testing.
Avinash
And some of you who are old enough uh probably like me have seen this journey and you can relate to what was happening in the early '80s, '90s or 2000.
Avinash
Uh testing was typically end of the life cycle phenomena, waterfall methods.
Avinash
Um some cases it was not given important um do itself.
Avinash
But we saw the transition.
Avinash
Though initially the transition was bit slow to to more automation um agile approaches, but things started changing in last decade when DevOps and continuous testing became the trend because every organization started moving to a tech digital first approach.
Avinash
So that cannot happen without DevOps and continuous testing or CI/CD approach.
Avinash
And that has kind of uh that helped everyone in the testing industry.
Avinash
So if you are a QA person, I think you require much more skill set.
Avinash
You got to learn different skill set.
Avinash
QA became very important in this phase.
Avinash
And that helped companies like us, product companies who are building test tools, dev tools.
Avinash
Um because in a continuous testing CI/CD kind of environment, you can't do anything without a tool like Pcloudy.
Avinash
But we are moving next step further into the future where testing is more testing is going to be more intelligent, autonomous and more collaborative.
Avinash
And the glimpse of it is what you will see today.
Avinash
So with that being said I'm sure you must have a lot of questions, right?
Avinash
You will have a lot of doubts.
Avinash
What AI can do, right?
Avinash
And a big question in everyone's mind is can AI really uh take over testing, right?
Avinash
Like any other field, can AI replace testing?
Avinash
Can AI replace humans in the testing world, etc.
Avinash
And I have found this slide capturing the answer very well.
Avinash
Um so if you look at the two sides of the slide, there are some low agency tasks and there are some high agency tasks.
Avinash
So low agency tasks are simple in nature, doesn't require complex environment, can be done as supervised, very more static.
Avinash
While the high agency tasks are very adaptive, it requires complex goals, complex environment.
Avinash
Now, you can see the this on the side of low agency uh the deterministic chatbots or LLM based agents are doing lot of wonders.
Avinash
Most of these kind of tasks can definitely be automated with agents coming into play.
Avinash
While the high agency tasks are still dependent on humans, right?
Avinash
And that's the current nature of it.
Avinash
And this gap I hope will always remain, but this gap is shrinking as as the day is progressing with the the fast changing world of AI and and GenAI especially.
Avinash
Now, I mean you you would have seen what happened 2 days back with SpaceX.
Avinash
Uh they they achieved something which was unthinkable 10 years back.
Avinash
Every scientist in the world said it's not possible, but they were able to catch the the the rocket booster midair.
Avinash
So never say no.
Avinash
There's nothing which is impossible.
Avinash
So I would believe uh in in never saying no.
Avinash
Uh I don't want to say anything is impossible, but I'm a big proponent of human in the loop.
Avinash
There are always tasks which humans will do while AI will support them.
Avinash
Of course, the technology change will make things much cheaper, faster, and there are tasks which was not possible beforehand in terms of automation.
Avinash
They would be automated, but still there are humans in the loop which will do the job for us.
Avinash
So, with that picture, I hope probably some of your doubts will get cleared.
Avinash
What's the What's the way forward?
Avinash
I think it's very clear that with with all the change happening, there's no other way than embracing the change, right?
Avinash
There will be naysayers who will doubt whether it's feasible, not feasible, but every technology at the beginning will have its naysayers and and and then it it gets adopted very very quickly, and that's what we are seeing in the world of AI as well.
Avinash
Um And and hear it from the the stalwarts, Satya Nadella again, if you don't jump on the new, you won't survive.
Avinash
And we don't want to be in this situation, and that's where I think the testing world has picked up very well in terms of adopting AI and bringing out very innovative solution with all the advancement which is happening with GenAI and GenTKAI.
Avinash
So, with that, I think Pcloudy also has been at the forefront of AI-driven software testing augmentation.
Avinash
And it's not something which we started just now.
Avinash
Uh we have a full AI stack where with lot of AI-based solutions.
Avinash
And in fact, our first solution, Certify, which was the exploratory bot, uh we released almost 5 years back when there was no GenAI itself.
Avinash
So, we had built this exploratory bot which can run test across different mobile devices on your app without writing a single line of code.
Avinash
It It has been used by many customers for monkey test, exploratory test, doing some sanity test across different devices.
Avinash
Then we released other AI utilities or tools within the stack.
Avinash
And two being very prominent are Q heal, which is a self-healing engine for any open source based script like Appium, Selenium, and other tools which can be used here.
Avinash
And the second one is Q lens, which is an independent visual testing utility which can again be integrated with any of your open source tools and frameworks that you are using.
Avinash
All this is also combined by our own GenAI GenAI based bot which we call it Alfie.
Avinash
So, Alfie is Alfie typically automates the task within the P cloud in platform.
Avinash
So, rather than navigating through the UI, you can say Alfie that go and find a device for me.
Avinash
It will find it for you.
Avinash
So, we have been kind of using this whole AI stack for quite some time and there are many other things in the pipeline.
Avinash
But today uh I want to focus on announcing the latest and greatest that we have built recently, Qpilot.
Avinash
And Qpilot is our platform agnostic autonomous agent for all things testing.
Avinash
And you'll see that in action and you'll see how it kind of makes your life easier, augments your testing.
Avinash
So, that's what we'll see it in action in some time from now.
Avinash
But let's let's look at slightly more Let's get slightly more deeper understanding of what Qpilot is both in terms of why we created, how it works, and what it does for you.
Avinash
So, in in a simple term, it's a agentic AI workflow to perform the tasks.
Avinash
And the task in the questions task in the question here is about uh testing tasks.
Avinash
Testing tasks across different platforms, be it your mobile application, web application, and we'll be adding APIs.
Avinash
So, it will be like end-to-end autonomous testing agent for different kind of testing use cases that you might be having.
Avinash
Now, you might be thinking what how is it different from GenAI?
Avinash
Because I can go and ask, let's say, ChatGPT to write a script for me.
Avinash
Uh but, this is quite different because we are using an agentic AI approach.
Avinash
And just to kind of give you a differentiation of how GenAI is different from agentic AI.
Avinash
Of course, agentic agentic AI is using GenAI at the back end, but the biggest difference here is uh the goal setting.
Avinash
So, agentic AI is goal-oriented.
Avinash
We'll You're giving a goal to the agent to go and perform.
Avinash
And agent then decides the the workflow, the orchestration of different agents, and finally achieves the goal for you.
Avinash
While GenAI uh is more uh either it can be a single-shot prompt or a multi-shot prompt, but you need to just ask a question, and the GenAI will respond.
Avinash
You don't know whether the the goal is being met or not, right?
Avinash
That's the main difference between GenAI and agentic AI.
Avinash
And to kind of give you slightly more detail of how this whole Qpilot workflow works, so you give a natural language input of your test case or test scenario, then it interprets and does the reasoning on its own by using different agents.
Avinash
Then it creates the workflow of how the script have to be script has to be created and then it it has the Once the workflow is created, it goes and execute the workflow and the final outcome is generated.
Avinash
And in between there are a lot of feedback loops between the workflow agents, execution agents, etc.
Avinash
So that's how any agentic AI framework works and that's what we have used for Qpilot as well.
Avinash
And again a different view, very simplistic view so that you get to understand what I'm talking about in the context of Qpilot.
Avinash
So at a high level there is a master agent which is a planning and orchestration orchestration agent.
Avinash
And then that agent opens up the workflow agents which are in this case like X flow locators generated generation agent, then the code generation agent.
Avinash
Once the code is generated it executes the step at the run time.
Avinash
And then if there are any error it analyzes and gives the feedback to the code generation agent.
Avinash
So it's all happening in real time where these agents are interacting with each other.
Avinash
And why this is much more important to understand is that our objective has been to give you a functional script of your scenarios that you write in natural language, right?
Avinash
There are many other I mean wrappers and solutions that you might see where they will give you script but which can work, which might not work.
Avinash
Here we take a bit of time through this whole agentic AI workflow to make sure that you get a fully functional script and you can even execute complex scenarios.
Avinash
And you will see that in action.
Avinash
So with that I think background it's important to understand that where does it fit for me, right?
Avinash
As a user where does it fit for me?
Avinash
What you see here is a typical life cycle of testing.
Avinash
So, you have requirement gathering, you define your test objective, design, test environment setup, create your test case, create CICD pipeline, execute, and the cycle goes on, right?
Avinash
Um, I I I need not explain this.
Avinash
You all have seen it day in day out.
Avinash
Uh, and this is where Qpilot Qpilot rightly, uh, fits in.
Avinash
So, you can create the script for you.
Avinash
Uh, it can execute the scripts across different environment.
Avinash
And when you when I say environment, it could be different devices, it could be different browsers, uh, it could be, uh, API execution.
Avinash
Uh, once the run once the execution is done, it can analyze the result for you.
Avinash
If there are some issues, it can take decision to fix.
Avinash
And, uh, then it can also run large regression suits for you.
Avinash
And another, uh, use case which is not mentioned here is synthetic monitoring.
Avinash
I mean, we are using Qpilot to run test 24/7 because it has the inbuilt self-healing engine as well.
Avinash
So, it it generates the script at the run time with the run time context of the application, and your script theoretically will never fail because it's taking generating the script at the run time.
Avinash
Uh, so so that's where, uh, Qpilot fits in.
Avinash
There are other areas as well where it can help, though we have not started, uh, delving into those area.
Avinash
For example, test design because I I personally believe that test design has to be done by humans.
Avinash
Uh, there can be augmentation, uh, within the test design process, but if the design is done well, I think rest of the process becomes very smooth.
Avinash
So, uh, requirement gathering, test design are should be better done by humans, and then rest of the task AI can help, right?
Avinash
That's the, uh, hypothesis which with which we have created this, uh, Qpilot.
Avinash
Now, uh, I hope I've given you enough context and and brief about what it is, why it exists, and where can it help.
Avinash
Um, I am excited to see it live in action.
Avinash
I'll hand it over to my colleague Shoaib who is the product specialist have been working with Qpilot board uh, quite long from the time we conceived.
Avinash
We have He has been showing demo to many customers, have been working with our POCs.
Avinash
So, he's the right person to kind of take walk you through.
Avinash
Uh, so Shoaib, over to you.
Avinash
Thank you so much.
Shoaib
Thank you so much Avinash for taking us through the journey of the AI and also the product from the people standpoint so where we are.
Shoaib
So, yeah.
Shoaib
So, without much delay, so let me please share my screen.
Shoaib
Avinash, if you can allow me to share my screen.
Avinash
Yeah.
Shoaib
All right, yeah.
Shoaib
Let me share it.
Shoaib
Right.
Shoaib
Uh, thank you so much again.
Shoaib
Uh, just quickly uh, give a a quick overview on what Pcloudy is.
Shoaib
So, uh, Pcloudy is basically uh, it's a cloud unified app testing cloud platform wherein it helps you test your different types of applications be it on the mobile side even across the uh, web automation side.
Shoaib
So, we provide a a comprehensive infrastructure in terms of providing the real physical devices as well as the browser instance where you can test your applications for your functional test and plus you can run your automation scripts by integrating with different automation solutions.
Shoaib
There is a seamless integration through APM, uh, Selenium, Espresso, XUI test, and so on.
Shoaib
And top of it, we have a new uh, you know, the automation suit which is more on the Qpilot side.
Shoaib
So, we will discuss more around it.
Shoaib
Uh, So, so this is a quick landing page where we have arrived.
Shoaib
So, once I log into the platform, so let me get into the devices section.
Shoaib
So, there are two different pieces here as I mentioned.
Shoaib
We help across, you know, test your applications on mobile as well as the browser instances where you can test your mobile web web application as well across various different OSs, OS combinations, different browser for your cross browser testing and so on.
Shoaib
So, let's get into the devices sections and talk about more on the devices piece.
Shoaib
This is a huge device infrastructure where we have real physical devices hosted across different part of the world.
Shoaib
And so, we support across various different OS versions, all the leading OS versions.
Shoaib
So, be it any version that comes in the market, we are the first to get the version and you know, have it on our mobile devices so that you can test your applications across different OSs.
Shoaib
So, we do provide all the leading manufacturers that comes across all the leading manufacturers that are available in the market across Apple, Google, Samsung and the latest models.
Shoaib
We don't just support smartphones, but also there's a support across all the leading smartphones, tablets and phablets wherein you can test your applications with wide different screen sizes.
Shoaib
At work, we also support testing your application for interruption based testing by providing a real SIM based devices.
Shoaib
So, if there are any scenarios for the interruption based testing like SMS or the calling functionality, you may pretty much test it.
Shoaib
And when it comes to the data center locations, so we are hosted across different part of the world.
Shoaib
We have few data centers in India, one in US and also in Singapore.
Shoaib
And we are moving towards the the Middle East and also falling with the Europe region.
Shoaib
So, that's something which is in our our road map.
Shoaib
Right?
Shoaib
So, that's about the device section.
Shoaib
So, now quickly let me uh, connect to any device.
Shoaib
For example, if I have any device in my mind, I can just quickly uh, connect to this device.
Shoaib
Yeah, so once I uh, connect to the device, the complete access of the device is provided, wherein I can, I know, test my applications in real time.
Shoaib
I can do the functional testing of my application.
Shoaib
I can uh, if my application is in the production, I can get it on the Play Store, App Store, download the app, uh, start testing my application.
Shoaib
As well as uh, while you're testing, we do provide certain key uh, parameters in terms of, you know, the device logs.
Shoaib
So, these uh, device logs helps you analyze if there any specific issues at the application side, or even any ANRs or the crash that happens while you test.
Shoaib
So, that's something which where uh, we will provide, which can help you analyze quickly on the issues.
Shoaib
On the performance metric side, uh, we do help you provide key performance parameters in terms of providing the uh, performance metrics like the CPU, the memory consumption, and the battery consumption, which can help you analyze if there are any performance bottlenecks within your app.
Shoaib
So, that's something uh, as a as part of our platform, we do provide these features, which can certainly enhance your uh, your functional testing experience.
Shoaib
And there are a lot many features.
Shoaib
I I will not deep dive into those.
Shoaib
Uh, we can always have a separate call, or we have some uh, readily available videos across all these different uh, features and functionalities, which you can uh, look forward.
Shoaib
So, now let's get into action, uh, talk about more about the uh, Qpilot.
Shoaib
So, now let me get into the Qpilot mode.
Shoaib
All right, so Qpilot is the first AI uh, based automation uh, mobile platform that Pcloudy has been launched, and we're the first in the market to support both mobile as well as web applications.
Shoaib
So, you can test your uh, different applications.
Shoaib
It's a uh, pretty much technology agnostic, so we support all the different types of applications, be it uh, React Native, be it Flutter based applications, or any hybrid and native applications.
Shoaib
So, pretty much support, and not just on the mobile, but also on the web web automation.
Shoaib
So, you may test your websites in real time uh, and uh, you know, automate the entire process.
Shoaib
So, let's take it take a look at the live how you can test your application.
Shoaib
So, I have a sample application that I have uploaded my cloud drive.
Shoaib
So, it's one of our e-commerce based applications where I can order certain products.
Shoaib
I can write the relevant application the app activity which is something which you can get it from your developers.
Shoaib
Let me define the name of it and the for the feature file I have just provided sample test case where I can just write the description in detail.
Shoaib
So, this description is something where you can capture it in simple English language.
Shoaib
It's a natural language, right?
Shoaib
So, which you provide as an input to the Qpilot.
Shoaib
So, then Qpilot takes into action.
Shoaib
It will start understanding the description that you have provided and then it will try to execute.
Shoaib
So, let's see it in live.
Shoaib
So, I'll just go ahead and then save it.
Shoaib
Maybe I'll just provide name to my test.
Shoaib
For example, test case one.
Shoaib
Save it and then I'll go ahead and then generate the script.
Shoaib
So, once I generate the script, now this is then you know, where in the AI the Qpilot comes into picture.
Shoaib
It analyzes the the application and also reads the descriptions whatever input that we have provided in simple English language.
Shoaib
It reads it and then gets in inside the DOM structure of the application.
Shoaib
It locates the element by using various different agents.
Shoaib
So, there are various different agents within the Qpilot as was mentioned in the in the slide.
Shoaib
So, it's we use we use the agentic AI approach.
Shoaib
So, which is intelligent enough understands the various different the the interpretations and the reasoning around it and it starts working.
Shoaib
So, while you're creating your automation script for the first time, so while you're generating it takes a little bit of time.
Shoaib
That's because at the back end it's getting inside your application, understanding the DOM structure, validating each and every description that you have provided with this within the script and it locates the expert and it does perform each and different action.
Shoaib
So, as you can see, it has launched the application and then you can also see the real-time real-time real-time update about the action that has been performed.
Shoaib
All right, so the first step has been completed where it went ahead and then clicked on allow option and now it's looking for the other option.
Shoaib
So, it will it will just go ahead and then perform all the discover user actions that you have been provided.
Shoaib
And yeah, and further I mean once the of course once the script gets generated, you can do a lot many customizations.
Shoaib
So, we will we will look into that once the script is generated.
Shoaib
So, we'll just go with the flow.
Shoaib
Let's let's wait for few more seconds and once the execution gets completed, we are good to talk about it.
Shoaib
All right, so it can be implemented as I mean as mentioned in earlier slides, right?
Shoaib
So, across different you know, within the different part of the within your testing cycle.
Shoaib
So, it can it will start by creating your test cases across web and mobile and you know, even you can analyze the test execution.
Shoaib
So, once the execution gets over, you can deep dive into your execution.
Shoaib
Once the execution gets over, you can look at look into the results how the impact has done.
Shoaib
What is the application?
Shoaib
I mean I mean what are all the different errors?
Shoaib
So, it does all that bit while you know, while the execution is happening.
Shoaib
And so, once the execution gets over, we will also show you talk about the reports how the report gets generated.
Shoaib
We will deep dive into the reports where it captures all the key metrics which talks about your test logs, the the device logs, the APM session logs along with the session with also all that will be captured which is part of analyzing your test results.
Shoaib
All right, so yeah.
Shoaib
So, we are almost there.
Shoaib
I mean we just few steps away.
Shoaib
Once this gets completed, we can get into the execution mode and see how we can execute this on uh know, multiple language support.
Shoaib
All right.
Shoaib
So, now I think we are pretty much there.
Shoaib
So, uh let me if I have a look at the steps where it is right now.
Shoaib
Yeah.
Shoaib
There's two more steps.
Shoaib
So, once these steps are completed, I think uh we are good to get into the execution mode.
Shoaib
So, it pretty much happens seamlessly.
Shoaib
So, as I mentioned, it's very much uh technological agnostic.
Shoaib
Be it any applications or be it any complex scenarios, uh it's intelligent enough to get inside your application and based on the interpretations and the reasonings and based on the models that we have been used.
Shoaib
So, it it it can help you uh you know, provide the scripts.
Shoaib
So, yeah.
Shoaib
So, the execution has been the the scripts generation has been completed successfully.
Shoaib
Now, I'll just let me switch to the Qpilot dashboard.
Shoaib
All right.
Shoaib
So, here is my dashboard.
Shoaib
So, so here there is a my suit different uh uh you know, the suits that I've already created.
Shoaib
I'll get me the one which I've created.
Shoaib
So, this is the one.
Shoaib
Yeah.
Shoaib
All right.
Shoaib
So, let's take a look at it.
Shoaib
Right.
Shoaib
So, feature name, the app details are provided and also the description.
Shoaib
And then if you get into in detail the generated scripts.
Shoaib
So, as you can see uh the code has been the the script has been generated uh flawlessly.
Shoaib
So, without any much effort even writing without a single line of code, I was able to just provide the description and capture the uh the code, right?
Shoaib
So, and we do support multi uh multi-language support.
Shoaib
So, as you can see, this is in the JavaScript, but we do also uh support in multiple language like even Python.
Shoaib
And we are going to add few more languages in our upcoming updates.
Shoaib
So, so, so look forward for you know, more uh language support that that keeps coming in our new updates.
Shoaib
And uh so, so the automation engineers or the QA engineers can deep dive understand uh the code.
Shoaib
If they want to copy this snippet, they can take it they can take it their existing uh IDEs.
Shoaib
They can understand they can just review the code.
Shoaib
Uh instead of writing the code, they can just uh review the code which is generated by the the Qpilot.
Shoaib
And then, you know, maybe if they want to add certain logics, all that can be done, but it's been taken care by the Qpilot itself.
Shoaib
And yeah, so as I said, it's in the multi language support.
Shoaib
So now, once the script is generated, you can go ahead and then execute it.
Shoaib
So let me just go ahead and then execute.
Shoaib
So I'll just trigger it.
Shoaib
Now, once I trigger, you will be notified that the execution has been begun and yeah, I can get into the live session now.
Shoaib
The execution has been begun.
Shoaib
So my active session is the place where you can actually monitor your live session, right?
Shoaib
So this will help you you can actually take a look at what really happening in the action, right?
Shoaib
So So once I get the they get the session, I'll be able to show it.
Shoaib
Yeah.
Shoaib
Yeah.
Shoaib
So it has connected.
Shoaib
So this is my session.
Shoaib
I can also get into the live action mode.
Shoaib
All right.
Shoaib
So now the device has been connected.
Shoaib
So now let's sit back and relax and see how the execution takes place.
Shoaib
I I can just sit and see the execution how it has been performing.
Shoaib
So yeah, so the execution has been started.
Shoaib
So execution will be quite faster.
Shoaib
I mean, only the script generation time time is a little more because of the the the the Qpilot and the various different interpretations and reasoning that we bring in.
Shoaib
So the execution will be flawless.
Shoaib
You can also execute on multiple devices at the same time.
Shoaib
So for example, if you want to execute the same script on multiple devices, so you can execute on different devices.
Shoaib
You can also choose the device on which you want to execute.
Shoaib
And not just that, we can also integrate this with a CI CD.
Shoaib
For example, if you're looking to integrate with your existing CI CD, so that's something which we will be integrating it very soon in our you know, couple of releases.
Shoaib
So we will have those updates very soon that's been added to our product.
Shoaib
All right.
Shoaib
So keep watch on on our updates.
Shoaib
So So that will have more enhancements towards the Qpilot.
Shoaib
So now the execution was was successful.
Shoaib
And yeah, once the execution is completed, you may get into the reports.
Shoaib
So, we do provide reports, uh, comprehensive reports, which will talk about, uh, all the detailed metrics where where we are capturing all the the test logs.
Shoaib
And so, this is the place where you can get inside the metrics.
Shoaib
You can get into the, uh, performance progressive report.
Shoaib
All right.
Shoaib
So, just for the benefit of the time, I have already, uh, I know shown one report.
Shoaib
So, wherein you can get to see the details of your, uh, build that you have just run.
Shoaib
So, it talks about the status of your execution, the the duration that it has taken and all the device details.
Shoaib
And then followed by, uh, all the key, uh, parameters, which includes logs of all the key capability key parameters in terms of it captures all the test logs, uh, device logs.
Shoaib
So, this is something, uh, it'll take some time, I know, while the device while the execution has just happened.
Shoaib
And also the APM logs.
Shoaib
So, this will help you analyze if there are any specific issues or any specific errors.
Shoaib
And you want to deep dive and understand what was the issue.
Shoaib
You So, you can get into it.
Shoaib
Uh, app filing.
Shoaib
So, this is more on, you know, providing the key performance metrics in terms of the CPU, memory, and the battery.
Shoaib
So, right now, uh, this is something, uh, which is there in our existing, uh, functional testing platform.
Shoaib
But on the Qpilot, this is something where we will we will go and update very soon.
Shoaib
So, this will be something where we will have our update in our upcoming, uh, releases.
Shoaib
And plus, we also have a a session video.
Shoaib
Session video will be, uh, taking you through all the detailed action that you have been performed.
Shoaib
Uh, the Qpilot has executed on the script.
Shoaib
So, that's something which you can have a look at it.
Shoaib
You may further download in case, uh, if you want to, you know, just download it on your on your machines.
Shoaib
And these reports are, you know, pretty much shareable.
Shoaib
So, you may share the, uh, report.
Shoaib
Uh, so, you can copy a share share you can create a link quick link that can be shared across different stakeholders, your, uh, you know, maybe different, uh, personas across different organizations, even with your leadership team, just to have a quick look about the, uh, report.
Shoaib
So, that's pretty much on the mobile automation side.
Shoaib
Maybe I can just quickly, uh, switch to uh, automation on the automation page.
Shoaib
So let me get into the mobile web automation.
Shoaib
Yeah, so this is more on the mobile automation side, how you can generate the scripts using Qpilot.
Shoaib
And now let's take a look at how you can generate your web automations in real time.
Shoaib
So let me take you through the web automation.
Shoaib
So we will connect one specific machine.
Shoaib
Now I'll let me take a Mac machine here.
Shoaib
I can just connect to any key browser machine.
Shoaib
So like physical real mobile devices, we do support you know for the web automations or even for the web testing scenarios.
Shoaib
So where you can run through all your websites, you can test your applications.
Shoaib
In case if it's a manual test, you can just log in, launch your application and then perform test across various different scenarios that you wanted to test.
Shoaib
You you may also sit with different resolutions in case of something where you want to see the behavior of your applications in real time.
Shoaib
Yeah, you may capture screenshots while you're testing.
Shoaib
So these are some of the capabilities from the manual testing standpoint that we already have.
Shoaib
So now let's get into the Qpilot mode.
Shoaib
So this is something which is of real of interest.
Shoaib
Yeah, I can just go ahead and then provide my details again.
Shoaib
I have a feature file already created.
Shoaib
Yeah, I can just upload this just for the benefit of the driver and of course the URL.
Shoaib
Yeah, so URL I save it as web automation.
Shoaib
Right, and now let me go ahead and then save it.
Shoaib
Yeah, maybe I just I can I can even create a suit in case if I am I want to create it and then provide some meaningful name to my test case, right?
Shoaib
One and then go ahead and then generate.
Shoaib
So once I generate the scripts again, now Qpilot will come into action, right?
Shoaib
So, uh yeah, it's just launch the application uh and then it will go ahead and then perform all the different actions that you have been written in the form of a natural language.
Shoaib
So, again, the the approach behind the scenes is pretty much the same that we have seen for the mobile.
Shoaib
Um we just wanted to show you even how it works for web automation.
Shoaib
So, now you may execute it across multiple platform as well.
Shoaib
You may you might have created the automation script on one platform.
Shoaib
Say for in this case, it's uh Mac machine.
Shoaib
Now, if I would like to execute the same script on the uh mobile automation on the uh Windows operating system, I can do that as well.
Shoaib
So, you don't have to change anything as such.
Shoaib
Just select the required uh OS and the browser that you want to execute and you can execute.
Shoaib
And plus additionally, you also have an option of uh you know, integrating with all the different uh the CI servers in case if you're triggering your build and you have your CI servers in place.
Shoaib
So, that's something uh which you can integrate and you know, test.
Shoaib
So, without any effort again, uh just imagine if I have to write a piece of a code.
Shoaib
So, there are a lot of dependencies again.
Shoaib
There's a dependency on right skills, source, uh right resources where the user has to understand the language and also writing creating the scripts, it's a time-consuming effort, right?
Shoaib
So, all these are I mean, something which can be which can be or which we can overcome and uh just create the scripts.
Shoaib
Even a non-technical user this is not to focus more only for the technical user.
Shoaib
Even a non-technical user can provide the steps in simple English and natural level language and then execute it.
Shoaib
So, the execution happens which you can see it on the screen.
Shoaib
So, it'll perform all the actions that you have generated and it'll take you through each each of the different steps, right?
Shoaib
It performs.
Shoaib
So, based on the uh the uh the agent AI approach, so it'll just go ahead, locate the uh objects, gets inside the DOM structure, uh gets the XPath relevant XPath and then I know, identify and then it validates.
Shoaib
All that can be done.
Shoaib
And uh further moving on, we will also uh help you uh know you know the from the customization standpoint.
Shoaib
You may wish to add certain assertions or validations.
Shoaib
So, all this can be done.
Shoaib
So, that's that's something which we'll be releasing in our upcoming releases.
Shoaib
So, watch out for the updates again.
Shoaib
Yeah, so it's just getting completed.
Shoaib
So, we'll wait for this to get over and then once this is there, I'll just quickly show you how you can execute it on the you know real browser machines.
Shoaib
Right, of course, as I mentioned, so we will also have a scheduler in place.
Shoaib
So, example for example, if you want to schedule it for a specific date and time or or specific on a specific Windows machine or a Mac machine with specific browser, so you can so we can always do that as well.
Shoaib
So, we want your support as well.
Shoaib
I mean, in this so you can test all your complex scenarios.
Shoaib
It's not just focus on one specific technology as I mentioned.
Shoaib
It's a technology agnostic, so you can test.
Shoaib
So, we are open to support you in this journey.
Shoaib
And, so look forward for all your participations and you know go through the trial of our on the Qpilot side.
Shoaib
Execute your with your scenarios and we can work with you and then help you out in this journey.
Shoaib
All right, I think we are pretty much there.
Shoaib
So, let me just check where we are.
Shoaib
Yeah, I think it's the last step one step that is pending.
Shoaib
So, it'll just add to cart and then Yeah, I think we are good.
Shoaib
So, So, yeah, I think it's completed.
Shoaib
So, as you can see the execution the script generation is pretty much done.
Shoaib
So, when this is done, you'll be notified with a pop-up wherein you can just get into the Qpilot dashboard.
Shoaib
So, let's get into the Qpilot dashboard.
Shoaib
Yeah, so there's the web automation.
Shoaib
Yeah, so here's my script.
Shoaib
All right, so it's generated.
Shoaib
Yeah, so script again you can just run through the script.
Shoaib
You can have deep dive.
Shoaib
Again, we have support for multi language even for the web automation.
Shoaib
It's not limited to any specific language.
Shoaib
We are going to add few more languages in our upcoming weeks.
Shoaib
All right.
Shoaib
So, yeah, I can just quickly have a review of my script.
Shoaib
Yeah, if everything is well or if I have to copy the code, I can just copy it and then execute it.
Shoaib
So, let's go ahead and then execute.
Shoaib
Yeah, I can just go ahead and execute either on Windows or Mac, as I mentioned earlier.
Shoaib
So, let me execute on a Mac.
Shoaib
Right, execution has been triggered.
Shoaib
Let's get into the live section.
Shoaib
Yeah, we'll wait for the um agent to pick up the machine.
Shoaib
So, once that's picked up, you'll be able to see a Yeah, I think it's picked a Mac machine on the Ventura OS version.
Shoaib
Yeah.
Shoaib
So, this will be seamless.
Shoaib
Also, the execution can be done.
Shoaib
I mean, you can you may execute it on multiple machines, multiple browser versions, OS different OS versions.
Shoaib
So, it it will be pretty much seamless.
Shoaib
And also, very important thing I want to talk about during his presentation is on the maintenance part of your test of your test scripts, right?
Shoaib
So, we often struggle to, you know, manage maintain the scripts.
Shoaib
What happens when there is a new build?
Shoaib
All right.
Shoaib
So, that's something, you know, where the the QA engineers try to put more effort, understand the change, and then create the scripts.
Shoaib
So, there's a lot of time, again, being spent.
Shoaib
So, with Qpilot, Qpilot comes with its inbuilt self-healing approach which automatically, you know, does your all your, you know, maintenance part.
Shoaib
So, it will it will just quickly regenerate the scripts and then you will be able to execute.
Shoaib
So, there's no there's it's a hassle-free from the maintenance standpoint.
Shoaib
You don't have to worry about the the changes which becomes in at the object property level.
Shoaib
So, it's something which which has been taken care.
Shoaib
And yeah, so now, as you can see, the execution pretty much just happened.
Shoaib
I mean, over the over the fly, I mean, we didn't see any any challenge at all.
Shoaib
So, I was able to execute.
Shoaib
So, likewise, I can execute it even on the Windows machines or any other machine that I want to look I want to execute.
Shoaib
Of course, CICD, even through CICD.
Shoaib
So, with that, I think I'm pretty much done with my demonstration.
Shoaib
And maybe I'll just quickly summarize what I have been done so far.
Shoaib
So, let me just bring out the slides.
Shoaib
Just to give a quick summarize about the Qpilot.
Shoaib
So, this is an, you know, where you can automate all your mobile apps or even your web applications, the websites autonomously.
Shoaib
You wherein you don't have to worry about taking, you know, or calculated effort.
Shoaib
It works on its own.
Shoaib
It It has an intel- It's intelligent enough that to to automate all your the all the complex scenarios.
Shoaib
So, you can test all your, you know, mobile apps, web apps, API execution can be done.
Shoaib
And we also spoke about the self-healing approach, which is very important, right?
Shoaib
So, more from the maintenance standpoint.
Shoaib
So, it's an inbuilt self-healing, which takes care of all the maintenance effort.
Shoaib
You don't have to worry about the changes in on your new build.
Shoaib
So, that's something which has been taken care.
Shoaib
And also, you know, going forward, we also have a lot of test orchestration.
Shoaib
You know, that will be for wider test coverage across browsers and devices.
Shoaib
So, it it's an intelligent enough to distribute your test.
Shoaib
So, you maybe you might have certain scenarios where you want to distribute your test across different devices or different browsers at the same time and thereby optimizing the test cycle here.
Shoaib
So, that's something where you can integrate.
Shoaib
And of course, as I mentioned CICD, you may integrate with all your existing CI servers that you've been using.
Shoaib
So, that's something which you can do it.
Shoaib
And of course, not but the least, so you can copy the script.
Shoaib
So, if you're an automation expert, you can just create the generate the script, copy the code, and then review it.
Shoaib
You can set it up in various different languages based on your existing frameworks and then work on it.
Shoaib
All right, so that's a quick summary about Qpilot.
Shoaib
And so we look forward for your active participation.
Shoaib
So if you want to try Qpilot, you can quickly scan your QR code from your mobile side from the mobile.
Shoaib
So this will give you access or you know, maybe you can quickly register to the Qpilot yes, so wherein we can help you provide the instance to you so that you can test all your complex scenarios, all your different mobile and web automations.
Shoaib
So we look forward for all your participation.
Shoaib
So with that, I think quick time check, I think we are pretty much there.
Shoaib
So we I'll open up for the Q&A.
Shoaib
Maybe Avinash, if you want to throw some some light on the on the demonstration or anything on the Qpilot.
Shoaib
Yeah, please you can go ahead.
Dinakar
So there are a few questions already um in the Q&A section.
Dinakar
The first is around um can I use the scripts generated within my automation framework?
Avinash
Um I think Shoaib just answered that question in the summary part and also while he was generating the script.
Avinash
Um that's one of the hypotheses that we have is uh it can augment the test creation for even uh test developers.
Avinash
So they can write the scenarios, uh generate the script, and copy within their frameworks.
Avinash
Uh however, uh for any other uh person who is just not interested in the script part, they can ignore that piece and just create their natural language scenarios and let Qpilot do everything.
Avinash
So that's an additional step which is there.
Avinash
Uh if you want to copy, you can copy and use it within the framework.
Dinakar
Thanks much for asking that question.
Dinakar
Maybe I'll just quickly take another question.
Dinakar
So, we have another question.
Dinakar
Can we work with Flutter based applications?
Avinash
Absolutely.
Avinash
So, as I mentioned, Qpilot is a technology agnostic.
Avinash
Be it any application, not just Flutter, but even the React Native applications on your even the native and the web the hybrid applications.
Avinash
So, we can pretty much automate.
Avinash
So, bring it on.
Avinash
I mean, test with with the trail instance that that does something which you can provide and you know, you can have a seamless experience.
Avinash
So, yeah.
Avinash
So, to answer to your question, yes, we do support the Flutter based applications.
Avinash
Oh, well, there are many questions that are coming up.
Dinakar
I'll just maybe just try to take the other questions as well.
Dinakar
So, how to provide test data in the Qpilot mode?
Dinakar
Very interesting interesting question again.
Avinash
So, so, when it comes to test data, right?
Avinash
So, funny one is we provide the test data while we are providing the you know, the test description.
Avinash
Right?
Avinash
So, you can provide all your test data.
Avinash
So, going forward, so we will also help you parameterize with different set of data.
Avinash
Wherein you have option wherein once the code is generated, you can go ahead and then provide the test data.
Avinash
So, you can map it.
Avinash
So, for example, if you have your data within your Excel and you want to pull the data out of it.
Avinash
So, that's something which you can do it.
Avinash
Right?
Avinash
So, that's something which you can pull the data directly from your Excel sheet and then you know, run through multiple set of data.
Avinash
So, test data is something which is crucial part of automation.
Avinash
Not just execute with the hardcoded values, but also execute with multiple set of data.
Avinash
That's a core feature.
Avinash
I mean, that's something which is expected from any automation platform.
Avinash
So, to answer to your question, yes, we you know, we we can support test data in the in the Qpilot.
Avinash
Hope that answers your question.
Dinakar
All right.
Dinakar
So, I have another question.
Dinakar
Uh is it only based on BDD framework?
Dinakar
I mean So, it's pretty much Yeah, maybe I'm not sure if you can I can take that question.
Avinash
Um Shivani, no.
Avinash
Uh what you saw uh was a uh totally natural language uh way of writing the test.
Avinash
However, we are not restricting the user.
Avinash
So, you can also uh write your test in a completely BDD kind of environment.
Avinash
So, if you already have your BDD uh format test scenarios like given then, you can pass that directly.
Avinash
In fact, we we are just uh working on linking that with the test management tool so that you don't have to manually copy-paste your scenarios in It will just directly fetch it from uh some of your uh BDD tools or test management tools.
Avinash
But, to to answer it specifically, no, it's not limited to BDD.
Avinash
You can write it in any natural language.
Avinash
In fact, it's expected to even understand the steps even if you don't provide detailed ones, but we don't recommend it because then there are chances of errors.
Avinash
So, you can all just say that, "Okay, can you uh test my uh flight journey within my website, right?" Uh it's it it is intelligent enough to understand, but there are chances of error.
Avinash
That's why we expect users to give at least the right steps in a in a natural language.
Avinash
Um […]
Avinash
Hope that answers your question.
Dinakar
Uh another question is um Uh all those asserts are yet to be introduced.
Dinakar
Can we enhance script manually and achieve […]
Avinash
Absolutely.
Avinash
That's the reason we have given the script.
Avinash
You can go and edit the script uh the way uh you want.
Avinash
You can add logic.
Avinash
You can add assertions.
Avinash
Uh but very soon we'll be adding the assertions uh itself within within a week or so.
Avinash
That That release is planned.
Avinash
Um […]
Dinakar
And a linked question is from Likith is how are dynamic locators handled after failure?
Avinash
Um so like I I uh kind of gave a hint in my diagram of the uh agent-based architecture.
Avinash
Uh so we have a error failure analysis agent which is uh continuously doing that.
Avinash
Every step execution it checks whether there is a failure.
Avinash
If If there is a failure, what could be the reason?
Avinash
Does retry.
Avinash
So it has the inbuilt uh self-healing mechanism.
Avinash
All right.
Avinash
Uh and uh so in that sense you should not worry about um running tests across different environments.
Avinash
And one of the beauty of it is and we have already seen the result um at many places.
Avinash
And let's say if you have two different instance of your application where there are some UI changes.
Avinash
For example, in Pcloudy itself we have two instances, one for public uh users which we call it public cloud.
Avinash
Then there is another instance which is a private cloud.
Avinash
Though most of the platform features remain same, there are some workflow related changes, there are login screen changes, um the landing page changes.
Avinash
In with Qpilot you can just write one script for these two because it's generating the script the run time uh for the login of whether it's a private instance or a public instance.
Avinash
So it can it can handle both the scenarios.
Avinash
Uh so it's minimizing all those challenges that we face across uh different environments, uh workflow changes, etc.
Dinakar
Uh there's another question from Amit.
Dinakar
How is the accuracy?
Dinakar
Suppose there are multiple objects with the same text level, how would you make it sure that it's working on the internet?
Avinash
That's the That's the core IP of it.
Avinash
We have built a lot of intelligence to make sure that it identifies uh even uh in a very complex uh kind of scenario.
Avinash
So, you see the uh the execution that we did was for Meesho.
Avinash
We We have tested it across Flipkart, Amazons.
Avinash
Uh we have tested it across MakeMyTrip.
Avinash
So, uh we have handled complex scenarios.
Avinash
However, uh we would definitely love to try out more scenarios and uh we we are continuously training models on new scenarios that we we are encountering.
Avinash
Um so, I'm not we can't claim 100% accuracy.
Avinash
Uh but it's much better than what we have seen uh anywhere else.
Avinash
So, I really hope that you try and uh give us feedback.
Avinash
And if your scenario doesn't work, there's a feedback uh uh button.
Avinash
You can just post it.
Avinash
Uh we'll reach out and and try to see where fails, and we'll uh make sure that the model learns from that failure.
Avinash
Uh […]
Dinakar
Uh sure, if you want to take uh questions from Anita.
Dinakar
Yeah.
Dinakar
Okay.
Dinakar
So, uh how can we uh how can we do send keys equivalent scripts through uh Qpilot?
Avinash
I think uh uh this question is more from the uh from the uh the uh the the main uh various different uh uh keys that we use uh you know, to uh send across while you're uh testing your applications.
Avinash
It's more of uh you know, providing the description and then uh and the Qpilot based on the uh the algorithms that we bring in the at the LLM models that has been prepared.
Avinash
Uh so, it's intelligent enough to understand all the key actions uh that you provide.
Avinash
So you can provide the keys in the form of a natural language again.
Avinash
Uh, so I'm just trying to uh visualize I mean what type of send keys it can be, right?
Avinash
Uh, maybe you know, if you can brief a little bit more around the what ex- what exactly that you're looking for or or we may reach out to you separately uh to understand your use case more in detail.
Avinash
And uh so that we can provide you the uh you know, the best possible uh solution answer to your question, right?
Avinash
So that you know, you can also try it out on uh on the Qpilot standpoint.
Avinash
But yeah, we can handle the Qpilot tends intelligent enough to handle all the various different uh combinations of task.
Avinash
Whatever that you provide, it's the it's all about how descriptive you provide.
Avinash
Right?
Avinash
So yeah.
Avinash
Hope that answers your question.
Avinash
Yeah, if there is any follow-up question, I mean I will help you answer.
Dinakar
Very interesting question from Surendranath.
Dinakar
Can I import back the edited updated scripts and would Qpilot be able to update the test cases from the updated scripts?
Avinash
Uh, very interesting use case, Surendra, and I think uh uh it it kind of throws interesting ideas.
Avinash
Uh, though we we have not kind of tested it yet.
Avinash
Uh, theoretically it should be possible.
Avinash
Uh, we'll try and see.
Avinash
Uh, but if it even if it doesn't work, I think it should be pos- possible for us to kind of bring that feature if that's a need.
Avinash
Uh, if I'll put it that way.
Dinakar
Yeah, thanks, Avinash.
Dinakar
And and there's a follow-up question with my Anit- I mean, I I think this is uh trying to elaborate more on the send keys action.
Dinakar
So like simple name and the password entry.
Shoaib
Of course, I mean, that can be done.
Shoaib
I don't see there is a there is a chance.
Avinash
Yeah, Shoaib, I'll just maybe add to it.
Avinash
I think uh I understand uh the scenario here.
Avinash
Uh so Anita, that is where I think the the agents are intelligent enough.
Avinash
So, for example, let's say tries with uh enter command and it doesn't work because it's executing it in the live mode.
Avinash
And if it's a failure, it will analyze the failure.
Avinash
If it is it is because of let's say the object could not be found or it is because the action could not be performed.
Avinash
It will retry with another kind of action.
Avinash
It could be send key.
Avinash
So, it has this intelligence built.
Avinash
So, we we hope that it's it works.
Avinash
Um we haven't specifically tried uh a send key scenario, but the way it has been designed, it should work.
Avinash
We can try the scenario on and let it let you know.
Avinash
And and in case if it doesn't work, we can just train the model and and it will start kind of taking it up.
Dinakar
Uh question from Amit, can it work on organizations' internal application which are not accessible over the internet?
Avinash
Um yes, it would.
Avinash
Uh but we'll require some white listing to be done.
Avinash
We have a private instance of the Qpilot which can work with internal applications.
Avinash
Uh but there there has to be some white listing where where our private instance can access your internal application, right?
Avinash
And that is how even our normal I mean our regular customers test their application on our private clouds, right?
Avinash
Either using VPNs or using the white listing approach.
Avinash
So, if that is done, I think it would work.
Dinakar
Um one one more question from Amit, is it possible to give the current automation framework as a context so that the script generation generation takes into consideration the existing page objects and variables.
Avinash
Uh that's a that's the next level of enhancements that we have been kind of working on.
Avinash
We have got this ask from the customers.
Avinash
Uh where they want to kind of use their existing framework as a context.
Avinash
Uh in fact, there has been asked where can we kind of modify those legacy frameworks uh into the into more latest uh approaches.
Avinash
So, these are certain new asks that we have got.
Avinash
Uh and I'm I'm sure that very soon you'll see some uh release and update around it.
Avinash
Uh but that's that's a very interesting uh thought and we have uh we have got similar kind of ask from some of the other customers as well.
Avinash
So, thanks Summit for sharing this.
Avinash
It validates the kind of ask uh and and gives more confidence that this is kind of use case we should also work on.
Dinakar
Interesting.
Dinakar
So, I think we have uh kind of reached uh a time limit here.
Dinakar
Uh if there are more questions, we'll be more than happy to answer.
Dinakar
Um thank you so much for uh joining and asking so many questions.
Avinash
What we need is uh you go ahead and try uh try as complex scenarios as you can and feel free to share.
Avinash
Uh um we can't we don't claim that everything will work for the first time, but we definitely want to work with you and support uh to make sure that uh we achieve the result together.
Avinash
Right?
Avinash
So, we are all excited to work with you uh during your trial, during your uh proof of concept.
Avinash
So, feel free to reach out and we'll be happy to help.
Avinash
Thank you.
Shoaib
Thank you so much, everyone.
Avinash
Thank you.
Dinakar
All right, great.
Dinakar
Thank you so much, Avinash.
Dinakar
Thank you, uh Shoaib, for sharing that insight.
Dinakar
And uh yes, please do try Qpilot.ai.
Dinakar
Um you've got the QR QR code.
Dinakar
And do reach out to us.
Dinakar
If you just go on our website, click on the chat box, uh you can just reach out to one of us, and we'll be happy to revert.
Dinakar
Or you can write to us at info@pcloudy.com.
Dinakar
See you.
Dinakar
Good day and good night.
