Dinakar
Okay, so welcome everyone and uh welcome to today's webinar on requirements to running tests in minutes.
Dinakar
Uh where AI agents we're going to learn about AI agents that write code and execute your entire test suite.
Dinakar
My name is Dinakar and I'll be the host for today.
Dinakar
And before we get started, I would like to go over a few housekeeping rules.
Dinakar
Uh you may have noticed that your line is currently muted.
Dinakar
If you have any questions, you can always submit them using the Q&A option uh given at the bottom of the screen.
Dinakar
Uh we will answer your questions.
Dinakar
We'll get to those questions towards the end of the webinar in the Q&A section.
Dinakar
And also please do note that this webinar is being recorded and we'll send out the link of the recording uh to share with your colleagues or watching it a second time.
Dinakar
Now with that said, let me introduce the speaker for today.
Dinakar
uh we have Jessica Selvin uh who will talk us through uh different aspects of how uh you know we can use the AI to write code and execute the entire test.
Dinakar
Jessica is a solutions engineer here at Pcloudy.
Dinakar
She has around four plus years of experience in the IT industry and prior to her role at Pcloudy she has worked with LTI Metree as a pre-sales consultant in automation solution design for various industries.
Dinakar
Uh she has an expertise in data analytics and various automation tools and of late she's been on the forefront of AI and you know testing technologies revolving around agentic AI.
Dinakar
So without further ado, let me hand it to Jessica to kick off this webinar.
Dinakar
Also Jessica, just to remind you that uh you're still on the presentation uh the presenter mode screen.
Dinakar
Uh I can Yeah, I can see your notes on the time and all of that.
Dinakar
So yeah, let's get that fixed before we get started.
Jessica
Yeah, let me just fix that.
Jessica
Just give me a second.
Dinakar
Awesome.
Dinakar
Yeah, you're you're good to go.
Dinakar
How about now?
Jessica
Uh, it's still coming up.
Jessica
Yeah, it's still coming up.
Dinakar
Okay.
Dinakar
Okay.
Dinakar
Let me share the other screen.
Dinakar
I guess if you're using multiple screens, they'll share the other one.
Dinakar
Um, yeah.
Jessica
Yeah, sure.
Jessica
Just give me a second.
Dinakar
No worries.
Dinakar
It's really nice to see a lot of uh people excited to listen.
Dinakar
I mean, we've got a full house today.
Dinakar
So, all the more excitement to hear from you, Jessica, about what you're going to share.
Dinakar
We're eager to listen from you.
Jessica
How about now?
Jessica
It's just loading and we're still on the presenter screen.
Jessica
Um,
Dinakar
do you want me to present and uh do you want to uh go?
Jessica
Yeah, that that also works.
Jessica
Yes.
Jessica
Yes, that works.
Dinakar
Okay, let me do that quickly then.
Dinakar
Let me just give me one second and I will share my screen.
Dinakar
Uh, let me pull up the I'm going to need a couple of minutes for that.
Dinakar
Just kidding.
Dinakar
quick couple of minutes.
Dinakar
Okay, I have the presentation here.
Dinakar
end the presentation will start.
Dinakar
I'm just going to download an offline copy so that it's easier for us to uh navigate through this.
Dinakar
Uh I'm sorry folks, we're having a slight technical glitch because uh I mean we're just trying to figure this out for the first time.
Dinakar
We we're using something new here.
Dinakar
So uh just bear with us for a couple minutes and then we'll be right with you.
Dinakar
So sorry for the technical glitches, but um we're sure to get started in a couple of […]
Dinakar
Okay, I'm just sharing my screen.
Dinakar
I hope you'll be able to see it.
Dinakar
Uh um share screen I think you should be able to see the screen right now.
Dinakar
Let me know if you're Yes, I can.
Dinakar
Um, yeah, thanks thanks for that and sorry guys for the uh technical glitch and uh taking up a couple of minutes there.
Dinakar
Uh, but yeah, let's just uh get started.
Dinakar
Uh the topic for today uh as Ducker introduced is how do you move from right you know your requirements to running your your test cases right and how does AI come into play both in generating your test cases and also your code and then the execution of your test cases right so uh yeah moving on to the next slide yeah so um what does the real world of test script creation look like today right and u let's look What do we see dayto-day beyond the ideal workflows because on paper automation u does look simple and in real projects however teams are facing very real and practical recurring challenges right so let's walk through some of the most uh common realities that we see today right firstly uh test creation time is time consuming right so even writing a single good uh end toend test can take hours especially when you're uh building for scale uh maybe across multiple devices or browsers environments right and it's not just about automating steps on a screen you have to deal with the test data setup uh right and the weight times and synchronization and there could be um environment instability uh right and um by the time a test becomes stable and reusable a significant amount of engineering effort has already been uh invested right So that's when the release uh time pressure comes in and test creation is often the first um activity that gets postponed.
Dinakar
Secondly, uh automation scripts can be fragile.
Dinakar
Uh tests break very easily in real applications.
Dinakar
Maybe a small UI change, dynamic locators uh right um or a synchronous load.
Dinakar
So even when teams are following the best practices, uh flakiness is still uh very much a real problem, right?
Dinakar
uh the application may be working perfectly fine uh but the automation fails and so instead of finding product defects teams end up debugging the uh test itself right and over time this creates a dangerous habit of uh where failures are assumed to be more of an automation issue rather than uh your application issue thirdly manual testing often lacks coverage uh this is not because manual testers are not um capable it's more of a scale problem uh right where uh there are too many combinations to cover such as different browsers and devices uh operating systems user role uh data variations and edge cases um as well as alternate flows right so with limited time in every sprint uh QA teams naturally focus more on um or maybe only the critical scenarios right and as a result edge cases remain um uh and you know the longtail user journeys often remain untested and the gaps slowly start uh showing and may even reflect in your production, right?
Dinakar
And fourthly, maintenance often eats up time and uh also your speed.
Dinakar
Uh most teams expect automation to improve speed, but sometimes regression cycles can become uh more of a bottleneck.
Dinakar
All right, a large part of the sprint is spent uh fixing broken locators or updating test data uh stabilizing the flaky runs.
Dinakar
uh right and in instead of expanding the coverage so the team is busy keeping the existing script alive um and then yeah because of which we see automation continuing to slow down uh but and as well as the real progress and finally uh test debt keeps growing.
Dinakar
So this is something that um is observed commonly across different teams right.
Dinakar
So uh new features in your application keeps getting shipped but automation cannot keep up with the pace of the development right automation is pushed to the next sprint and then to the sprint after that.
Dinakar
So very soon there is a growing backlog of um maybe unautomated features partially maintained script and uh unstable regression flows.
Dinakar
So this is what we call as the test debt and just like technical debt it quietly increases with every release.
Dinakar
So just taking a very simple real world example right.
Dinakar
So say in one sprint a team releases a new feature called uh a rewards wallet right uh for a fintech app uh and manual QA validates the feature but automation is delayed and in the next sprint maybe a small UI change is introduced for the same user flow.
Dinakar
So now you have two problems that are coming in at the same time, right?
Dinakar
Firstly, uh the existing automation breaks and then needs to be fixed and secondly the rewards wallet feature still needs to be automated.
Dinakar
So the team is now working on old scripts as well as new scenarios both in parallel.
Dinakar
Uh and this is how regression debt is created, right?
Dinakar
Because not because teams are careless but because test creation and maintenance cannot keep up with the speed of development.
Dinakar
So when you put these five uh realities together right uh the time consuming test creation or fragile automation uh limited coverage and high maintenance effort right the pattern becomes clear so it's not just a tooling problem or a framework uh issue right the real challenge is that the traditional way of creating and maintaining test scripts does not scale um according to how fast we're seeing applications uh change right and that's part of that gap is what we're going to be uh focusing on uh during this session.
Dinakar
Next slide.
Dinakar
So um yeah, let's look at the what the core problem is, right?
Dinakar
Um in terms of structure and scale.
Dinakar
So if you look at how teams still create tests today, test coverage is still directly proportional to two very simple things.
Dinakar
uh the number of testers you have in your team and the amount of time you can spend.
Dinakar
Uh so in other words, if you want more coverage, the only practical way to get it is either you add more people to the team or you free up uh bandwidth for the team, right?
Dinakar
So that that's already a limitation.
Dinakar
But at the same time, the environment we're working in is moving in the opposite direction.
Dinakar
Meanwhile uh because the product velocity is increasing uh right sprint uh cycles are shorter teams are releasing more frequently uh right and u the complexity of apps are also increasing be it especially in BFSI domains or healthcare and insurance uh right where every flow has compliance requirements as well as validation rules um and also the device fragmentation is increasing right with more uh devices OS versions screen sizes uh right so coverage is still constrained by people and time and yet even with all this tooling um the frameworks we have today for test creation itself is largely manual uh right and so that's one of the reasons why it doesn't scale right so let's quickly walk through how test scripts are typically created uh you know in most teams is uh yeah mainly manually uh right through gathering requirements and uh uh manual efforts that highly limit how you can scale uh your test coverage.
Dinakar
So let's go to the next slide to look into that more in detail.
Dinakar
Yeah.
Dinakar
So um firstly, yeah, as I mentioned, let's look at the code problem, right?
Dinakar
And how our teams actually scaling up their test cases and test coverage, right?
Dinakar
So 20 to 30 hours can be spent on just writing your manual test cases for any typical feature or uh user flow teams can easily spend multiple days just creating the test cases as well as the test scenarios.
Dinakar
Uh right so this includes understanding the flow identifying the validations uh preparing and reviewing the uh cases with stakeholders.
Dinakar
So even before automation starts a large chunk of the sprint has already been consumed right.
Dinakar
Then secondly, additional time is spent on your automation scripting.
Dinakar
Right?
Dinakar
So once the test cases are ready, the same flows have to be translated again into automation code be whichever framework you're using, selenium, APM, playwright.
Dinakar
Right?
Dinakar
So again, this is a manual engineering activity that consumes time where you need uh also skilled resources and framework Thirdly, uh requirements that change midway, right?
Dinakar
So while test cases and scripts are being prepared uh requirements evolved right UA UI changes can happen or maybe the acceptance criteria has changed right so this immediately invalidates the part of the work already done so there could be cycles of rework uh right test cases need to be updated uh automation scripts also might have to be reworked on all right and fourthly production bugs still uh end up existing uh despite all this effort defects will escape into production.
Dinakar
Again, not because our teams are careless, but because coverage cannot keep up with the uh you know the rate of speed of change that's happening.
Dinakar
Uh right and edge cases may have been missed.
Dinakar
Uh right and because the regression cycles uh were constrained by time.
Dinakar
uh releases also can get delayed.
Dinakar
All right, testing becomes a gate that slows down uh delivery where scripts are unstable or coverage is incomplete and the automation backlog uh continues to grow.
Dinakar
So more features get delivered every sprint but the capacity of automation remains uh limited.
Dinakar
So yeah, so that is the in in summary here are all the challenges that are facing uh that uh you know QA teams face that block becomes a bottleneck in terms of being able to achieve good coverage.
Dinakar
Moving to the next slide.
Dinakar
So testing has entered the age of AI agent.
Dinakar
So let's talk about this shift right in light of the problems that we uh discussed.
Dinakar
How will AI agents can it really make a difference and how does it help solve these uh problems?
Dinakar
So firstly um AI agents are proactive rather than reactive.
Dinakar
So in traditional testing automation is reactive.
Dinakar
It waits for the commands.
Dinakar
Uh we explicitly tell uh what step to run next uh right in what order what are the conditions.
Dinakar
So if we do not script it, it simply will not happen.
Dinakar
Right?
Dinakar
And with AI agents, that behavior changes fundamentally, right?
Dinakar
Uh agents are more proactive.
Dinakar
They can take initiative based on context and signals from the system.
Dinakar
Uh right, instead of waiting for hardcoded instruction.
Dinakar
So agents can decide to explore a flow, retry a scenario maybe in a different way.
Dinakar
Uh right.
Dinakar
So it's not just about faster execution but it's more about moving uh from a system that responds to instructions to one that actively participates in testing.
Dinakar
Secondly we also have um task focused versus uh being goal oriented right so AI is uh goal oriented.
Dinakar
So for example uh in a simple way to put it right in traditional automation we can create scripts such as test login right and there is a specified flow uh right so each script is tightly coupled to a specific path and a UI flow right but in real quality is not about completing a task it's about achieving an outcome right have I for example the goal could be ensuring secure access so you'll be able to uh test whether your AI agents instead of telling the system test login is it actually able to ensure uh secure access to the application right so you can it'll be able to understand your intent rather than just specific tasks uh thirdly um collaborative versus um being working in isolation so when you're having an AI system in place for your testing um one test does not know what are the test covered right the areas would be uh unstable which areas are unstable uh right and AI agents on the other hand can coordinate with other agents and systems right and fourthly and finally we have static rules versus learning and uh evolving strategy so um in traditional automation it's static right it follows rules that were created at the time the script was created so when the UI changes uh the flows change and then the timing behavior [clears throat] changes and hence the script break right so AI agents are designed to learn they observe what fails and where instability uh appears right so how application behavior can change over time and yeah that's what we're talking about and that's what when we uh we mean when we say AI is uh non-deterministic uh by design right and it breaks down human dependence and isolation so moving to the next slide yeah So again, I just want to uh you know, let's take a pause and reframe how we think about the future of testing, right?
Dinakar
Because we're moving into a world where uh AI agents are fundamentally changing how we work with technology.
Dinakar
Um there is a yeah there's a quote here by a leader in the AI space, right?
Dinakar
So artificial intelligence is the new electricity and how it transformed countless industries.
Dinakar
AI will now do the same.
Dinakar
Right?
Dinakar
So the reason this comparison matters is uh not because u AI would just simply make existing processes faster.
Dinakar
It will completely reshape how industries are uh designed, how work is done.
Dinakar
Uh right.
Dinakar
And that is not just a productivity upgrade for testing.
Dinakar
it uh represents more of a structural shift in how quality engineering itself will evolve.
Dinakar
Right?
Dinakar
So AI agents are not just models that generate text or write snippets of automation.
Dinakar
Uh it's more like you can see it as something that operates on our behalf uh that can understand your context uh goals, what are the constraints and continuously act and adapt accordingly as well.
Dinakar
So this is what um this is how AI can actually um cause a m not just a technical shift but a mindset shift in how [clears throat] teams are collaborating with technology.
Dinakar
So instead of tools waiting for instructions we begin working alongside these systems.
Dinakar
Uh right so uh yeah this is another quote uh AI agents will become the primary way we interact with computers in the future.
Dinakar
Right?
Dinakar
So they will be able to understand our needs.
Dinakar
um and preferences and proactively help us with tasks and decision making.
Dinakar
Right?
Dinakar
So that's the shift um that we need to get into uh in order to understand these new changes that are happening in the QA space.
Dinakar
So moving on to the next slide.
Dinakar
Yeah.
Dinakar
So the problem with uh yeah this is just to show what is the inflection point in terms of how are human uh uh dependent workflows happening uh versus agent-driven uh workflows right so um when we start from requirements a tester in [clears throat] uh interprets the requirements test cases are written and the automation engineers convert them into scripts all right and quality will end up depending more on an individual effort rather than a standardized form of s you know system intelligence right so that is the biggest challenge right because you're not able to standardize the quality of the test cases and it depends more on individual uh effort right and because of that you will not be able to scale such a system and there will always be some gaps uh when you are generating test cases like that now coming to how AI agents u actually change this right so in the agent-driven model uh requirements directly go to an AI agent, right?
Dinakar
The agent understands the intent uh derives structure test suites uh generates executable code and continuously adapts as the system uh evolve.
Dinakar
So this will what is this is what will create uh shift left testing right so testing that starts earlier and uh evolves continuously.
Dinakar
Moving to the next slide.
Dinakar
Yeah.
Dinakar
So with that we would like to introduce our QGEN agent for the first time.
Dinakar
So this is our platform agnostic autonomous AI agent uh designed to intelligently generate test cases directly from requirements uh helping teams move faster improve coverage and embed quality earlier in the delivery life cycle.
Dinakar
All right so that is just a short definition.
Dinakar
Of course, we will have a live demo of this and we will be able to see that more in detail.
Dinakar
And uh moving to the next slide, let's just see what are the key problems it solves, right?
Dinakar
Um so firstly, no more manual test case uh drafting.
Dinakar
Uh right, you don't need to start from a blank page.
Dinakar
So instead uh of testers manually analyzing every requirement uh or document or user stories, right?
Dinakar
uh it is able to understand the intent of the requirement and automatically generates the meaningful uh structured test cases and secondly multiple format inputs.
Dinakar
So one of the biggest challenges in real projects is that requirements are never stored in just uh one format.
Dinakar
Right?
Dinakar
Some teams work with user stories in Jiraa.
Dinakar
Other uh maybe they have product flows as documents or PDFs and others rely on design files or spreadsheets.
Dinakar
Uh right so QGEN is worked to um is designed to work exactly in this reality.
Dinakar
Right.
Dinakar
So it can ingest multiple um requirements from in different formats and sources and it's of course powered by agentic AI.
Dinakar
Uh so it's not just a prompt based automation tool.
Dinakar
Uh it can reason, plan and also iteratively think and finally it's platform uh aware.
Dinakar
All right.
Dinakar
So another critical challenge today is that test cases are often written in a platform agnostic way and you need need to later rework it.
Dinakar
Right.
Dinakar
So uh QGEN is platform aware by design.
Dinakar
Uh it supports web applications, mobile apps and also API testing.
Dinakar
Uh right and both Android as well as iOS.
Dinakar
So it's able to cover all these different platforms in a single um uh agent.
Dinakar
Right.
Dinakar
So with that let's get into the demo.
Jessica
Um thanks to Dakar for sharing the screen.
Jessica
Maybe I'll take over from here.
Jessica
Is my screen visible?
Dinakar
Uh, not yet.
Dinakar
Uh is it visible now?
Dinakar
Yeah, we can see your screen.
Dinakar
Yeah, we can see your screen.
Dinakar
Uh we are not able to see you though.
Dinakar
Uh but I mean go […]
Dinakar
All right.
Dinakar
Yeah.
Dinakar
Awesome.
Dinakar
We can see
Jessica
Yes.
Jessica
So um yeah, this is introducing our Pcloudy platform firstly.
Jessica
Uh right.
Jessica
So uh we are a SASbased platform.
Jessica
Uh right.
Jessica
And the core of our solution is that we give you um our test infrastructure where we host uh real mobile devices and browsers for you to um launch your apps and do your testing.
Jessica
Right?
Jessica
So that's how we started.
Jessica
Uh and that's the core of our solution.
Jessica
However, on top of that, we have built several AI solutions.
Jessica
Uh right, so all these devices that you see here are uh real devices both on Android as well as on iOS.
Jessica
And uh we also host browsers for you to do your browser based testing.
Jessica
So uh that's just a quick um you know overview on in terms of our test infrastructure.
Jessica
With that let me get into the focus of what our demo is about today which is our QGN agent.
Jessica
Yeah.
Jessica
So you can create your projects and have a folder structure here.
Jessica
Right.
Jessica
So you can create multiple projects.
Jessica
So I have multiple folders created for different apps.
Jessica
So today maybe we will uh run a quick test scenario uh generation for the IKEA mobile app.
Jessica
So let me just select and create a new test case.
Jessica
So again you will have an option to select between mobile and web.
Jessica
If it's mobile then whether it's Android or iOS you can give a test case name and a description.
Jessica
So I can just maybe so maybe the the scenario we are testing for is test results uh sorry search results right so and it's an Android based application.
Jessica
So you can also give a description.
Jessica
So I'm just going ahead and going to the next step.
Jessica
So here again as I mentioned we have multiple ways of giving um your inputs right.
Jessica
So you can I mean soon we will have the Figma URL also supported and you can directly upload your APK or IPAs as well.
Jessica
Uh you can also give feature descriptions or any of your PRD documents or um uh user stories.
Jessica
All right.
Jessica
So you can give it in the form of a documentation or you can go ahead and directly drop screenshots maybe of the Figma design or directly of the application.
Jessica
All right.
Jessica
So I'm just um using the screenshot option here and one of the screenshots.
Jessica
You can add multiple screenshots.
Jessica
So here I'm just selecting one and I can go ahead and click on analyze.
Jessica
Right?
Jessica
So firstly what is happening is ingesting the content that you've fed be it the PRD documents, the Figma, uh the screenshots, right?
Jessica
And you can give multiple types of formats as well.
Jessica
It's not just one.
Jessica
And then it's analyzing the different scenarios post which you will be able to shortlist and see the different test cases that are generated.
Jessica
Right?
Jessica
So yeah you can see the analysis is now complete.
Jessica
It's given you uh eight suggested flows and you can see all the different scenarios here.
Jessica
Right?
Jessica
So these are not the test cases.
Jessica
It's just giving the recommended scenarios from which it will generate the test cases.
Jessica
So uh you can filter out what you want.
Jessica
uh so for example I can just deselect all and just pick out the few ones that are uh relevant to me.
Jessica
So there is very much a human in the loop um still present because uh ultimately uh the decision uh should come from uh the QA engineers in terms of the relevance right so we very much uh have the human in the loop so it's not just the final set of test cases that are going to be determined for you so you can pick out which are the relevant scenarios and if you feel that uh maybe I need multiple iterations I need uh more edge cases for example or I need more um uh invalid uh input or edge cases or error handling or validations right you we have an option to uh refine with AI right so you can choose uh this option here and you can add another attachment if it's uh relevant in case or you can just simply give a feedback right for example add more edge cases right or something like right and then you can submit and update the flow as well right so it will submit that and give me all the edge cases as well.
Jessica
Right?
Jessica
So you can it will you can give multiple feedback and iterations uh till you're satisfied with the u you know the final um set of scenarios.
Jessica
So here it's given me more uh edge cases as well.
Jessica
So now let's just go ahead and uh generate this.
Jessica
So now during generation it might take maybe 30 seconds or so to uh uh generate uh your test cases.
Jessica
So basically what's happening is based on the scenarios you have selected it is going to give you the full uh set of test steps as well as u the different test cases that are uh […]
Jessica
So meanwhile maybe I can quickly show you a completed one.
Jessica
So yeah this is how a final output would look like.
Jessica
So based on the different scenarios you have selected, you can see uh the test cases that are generated uh for the application.
Jessica
So you can see it all numbered.
Jessica
Again, you can choose uh in case if something is irrelevant, you can choose or you can even edit it out before you finally export it in form of a CSV file or an Excel file.
Jessica
So you can see all these are the different scenarios created.
Jessica
You can uh here it's created about 14 scenarios uh test cases or so.
Jessica
All right.
Jessica
So you can see how um you know there's no limit as such and the more screenshots or PRD documents you add the more extensive your uh test case uh outputs would be right and I can also choose to export it uh as a CSV file right so you can do that as well.
Jessica
So let me just open up how that would look like.
Jessica
So this is just a sample document to show you how uh your extracted final results will be in the Excel where you will get the scenario ID as well as the test case ID uh descriptions what are the prerequisites uh right and uh the test case preconditions and finally the test the detailed steps of the uh test case flow itself right so you will get all of that and you can export it as an excel or CSV or soon we will also have a link to our Qpilot AI agent.
Jessica
So you can directly convert this into automation scripts.
Jessica
By taking these steps, it will feed it into the app and generate the final ready to run script and then you can choose to run it on the PCR uh devices and browsers.
Jessica
Right?
Jessica
So that's how it'll become an end toend uh autonomous agent for you.
Jessica
So um yeah, that's on our QGN agent, right?
Jessica
So let's quickly see.
Jessica
Yeah, even the one that we just now created the test cases have been generated as well for this.
Jessica
So you can see all the edge cases that we had asked through the AI refining feature has been uh generated.
Jessica
So now that is on QGEN uh right so I want to quickly cover Qpilot as well because together [clears throat] they work as one agent uh right so whatever manual test steps you have uh gotten from there uh that's how you can create full end toend test cases um here in our Qpilot agent so let me just create another sample test case maybe I'll select a sample app, give a test case name and the package and activity name.
Jessica
Right?
Jessica
So this will help me um help the Q- pilot uh model uh to know the um which app to launch onto the device.
Jessica
I'm just selecting a device and going to the playground.
Jessica
So right now what's happening is firstly the device the Samsung device that we had selected is going to get connected to and launched and then the uh sample app the APK that we had selected along with the package and activity name will uh help it launch uh the application onto the device and then thirdly the Qpilot model is also getting initiated at this time right so that you can give natural language inputs and it will be able to uh generate the script.
Jessica
So now that's happened uh you can see the app is now uh launched onto the device and now I can just start giving plain English language descriptions like for example enter the username in the username Good.
Jessica
Right.
Jessica
So what is happening is each step by step what is happening is it's understanding what is the intent of the user what is the closest match and it'll do that particular action step by step.
Jessica
Right?
Jessica
So you can see here that uh the expath locator has been uh generated and also it's reasoned.
Jessica
Now I've also selected few more steps to add uh like the entire login flow.
Jessica
So let's just give those steps as well.
Jessica
So all that will be cued and it'll pick up uh one after the other.
Jessica
Right?
Jessica
So what is happening is that it's not a keyword based or a typical record and playback kind of a solution.
Jessica
Right?
Jessica
So here uh the Qpilot model is understanding and going through the entire DOM structure seeing what is the closest match and it'll give you the reasoning as to why it selected that particular element in the DOM structure.
Jessica
Right?
Jessica
So you can also add validation steps.
Jessica
So for example validate if the login button is enabled.
Jessica
Uh right or validate if so and so text or popup is present.
Jessica
All right.
Jessica
You can give if conditions you can capture screenshots.
Jessica
You can give um wait times.
Jessica
Right?
Jessica
So all of this um or slide actions uh right all of this is possible uh in the Qpilot u model right so that's how you will be able to the manual test cases that you've generated you will be able to convert them into automation scripts so uh throughout the entire process there is no coding involved and you don't need any specific automation or framework uh skills right so I've just created a small flow uh but there is no limit to the number of uh steps you can uh add on here and you can complete your entire flow.
Jessica
And once that's done, you can go ahead and finish your test case.
Jessica
I'm finishing it here.
Jessica
So this is the test case we have just now created.
Jessica
Coming to the code, you can see the entire ready to run code is generated here.
Jessica
Right?
Jessica
So you can see all the uh locator information is over here.
Jessica
And you don't need to look into the code though we do give you access.
Jessica
You can copy it and modify it in your own frameworks.
Jessica
We do give that option.
Jessica
Um but uh you don't need to um debug this uh right because already it has been validated while uh generating itself.
Jessica
Each step has been validated.
Jessica
Right?
Jessica
So I can just simply go ahead and choose to run the script um on any number of devices.
Jessica
Right?
Jessica
So I can just select uh run test and it will go ahead and execute it on the devices.
Jessica
So any uh automation that is getting executed can be tracked in the my active sessions here.
Jessica
So you can see the device that we've selected [clears throat] has already been um initiated.
Jessica
Now once it starts to uh the ongoing process I'll be able to see the live view here.
Jessica
Yeah.
Jessica
So the execution has now completed.
Jessica
So I can come to the report section here.
Jessica
And as you can see uh the entire automation flows uh or your test builds all of them can be tracked in our um AI analysis report.
Jessica
So you can see this this is the automation I've just now run.
Jessica
So you can see that it's in the past state.
Jessica
I can see the text logs which is the output of each and every step.
Jessica
And you will also have the video of the entire session.
Jessica
And in a few seconds even the logs will get populated here.
Jessica
And you will also get AI based analysis.
Jessica
Um right.
Jessica
So that'll take a few seconds for the entire logs to uh generate and you can also log a bug from here.
Jessica
So we have integrations with Jira and Slack.
Jessica
So you'll be able to in case if there's any test failure you want to directly uh log it into your Jira you can do it from here itself and all these reports are sharable as well.
Jessica
So you can click on share link and you will be able to uh share it to even users who are not registered on the Pcloudy platform uh will be able to view the results of your test cases.
Jessica
So that is uh all I wanted to show in terms of our end to-end flow of how QGEN uh is able to write from screenshots give you full end toend test case scenarios and test cases and how that can be converted into automation scripts and you can run them on real devices and get the results all on one platform.
Jessica
So I hope uh that was informative for you and uh it sparked some thoughts and ideas uh for you to try on your own as well.
Jessica
Um feel free to reach out to us if you want maybe a trial or if you are um interested to use this for your own teams.
Dinakar
Uh right and we'll be happy to take any questions you have now.
Dinakar
Uh I think we have a few questions uh on our uh Q&A section.
Dinakar
Um, not sure if you're able to view them, but I can read them out if you'd like.
Dinakar
I think the questions is Okay, great.
Dinakar
Great.
Dinakar
I'll just read out the question and then maybe you can answer or I can read it for you and then you can answer whatever works.
Jessica
Okay.
Jessica
[clears throat] Question is what is
Dinakar
Go ahead.
Jessica
Yeah.
Jessica
Yeah.
Jessica
Yeah.
Jessica
So, Sang has asked what is the full form of QGEN?
Jessica
Uh, right.
Jessica
So, Q a lot of our agents like QG gen, Qpilot, Qlens.
Jessica
Um, Q stands for quality.
Jessica
All right.
Jessica
And gen because it generates test cases, right?
Jessica
So, that is the reasoning behind naming it as QGEN.
Jessica
Uh, yeah.
Jessica
Then Sham has asked, you just created test cases.
Jessica
Can it not automatically take from there and run as automated tests?
Jessica
Yes.
Jessica
Right.
Jessica
So as you saw um that is a feature that will soon be live.
Jessica
All right.
Jessica
So you can directly export that and uh ask it to convert it and through our Q- pilot agent which I just now demonstrated.
Jessica
Right?
Jessica
It will be it will automatically convert it uh to a fully working script.
Jessica
Right now you might have to just copy those steps and paste it into your our Qpilot agent but soon that will be a seamless uh process.
Jessica
So Ashbad has asked will the agent help in uh preparing new test data specific um to this test scenario suggested right because test data would not always be available.
Jessica
So yeah, it will suggest the test data as well, right?
Jessica
So and if you feel that maybe it has not sufficiently suggested that um the refine with AI option is there where you can ask it uh to specify uh the test data for the edge […]
Jessica
Uh Chetra has asked will it be helpful only for mobile automation?
Jessica
Uh no right we support web equally as well.
Jessica
I have just demonstrated it on uh mobile but uh you can simply give your URLs and it works exactly the same way even for web.
Jessica
Uh again uh Sangana has asked which LLMs are used.
Jessica
So we do use enterprise LLM right uh enterprise licenses for uh LLMs that are being used uh for behind the scenes both for QGEN and Qpilot.
Jessica
Um and also on API testing, yes, we are supporting API tests as well.
Jessica
That will soon be live where and even right now it's already live where you can add API steps into your Qpilot uh and it will be able to give you the outputs of that.
Jessica
Is QGEN open to everyone?
Jessica
Uh yeah.
Jessica
Uh so we'll be happy to support in any form of trial that access that you want for a day or two and uh we can best for you.
Jessica
Can we integrate this with existing infra?
Jessica
Uh yeah we can have a separate discussion on this on what exactly is your use case.
Jessica
Uh right and we can help you.
Jessica
We are open to customizations.
Jessica
uh right in cases where um you know we see that as a requirement for you we will be uh happy to look into this.
Jessica
Uh can we automate API and DB tests?
Jessica
Uh DB tests maybe um in the future but uh API tests yes you will you will be able to do
Dinakar
Yeah.
Dinakar
So I think I've covered all the […]
Dinakar
Okay.
Dinakar
I think there is one more.
Dinakar
Yeah.
Dinakar
Yes.
Dinakar
Yes.
Dinakar
If we have large multiple page websites, are we supposed to upload the screenshots?
Jessica
Yeah.
Jessica
Uh you can upload multiple screens at the same time.
Jessica
All right.
Jessica
Or you can give the PD document uh or a Figma URL.
Jessica
All right.
Jessica
So all these will work and you can upload multiple screenshots uh to generate the screenshot of an end to end flow say that involves multiple screens.
Jessica
Right?
Jessica
uh you can uh upload all those screens and it will be able to connect that and get uh give you logical test cases based on that.
Jessica
So I will not be able to name the exact LLM being used.
Jessica
Uh we can we do uh maybe we can have a separate discussion if you want more information but uh as I mentioned again they are enterprise
Dinakar
Yeah.
Dinakar
Uh um I'm just going to read out those questions so that our audience knows what we're talking about.
Dinakar
Uh so one of the questions that uh Jessica just answered was this question about which LLMs are used in the QGEN platform.
Dinakar
Uh so that is what Jessica just answered and the other one about uh the screenshots of the website was if we have a large multiple page website are we supposed to upload the screenshots of all screens to generate the test cases and like Jessica said you just need to yeah if you have a large you can either upload your Figma where you know where designers design the website and put that link in there or if you need the entire complete flow just put in all the screenshots there and the the AI agent is intelligent enough to kind of connect the dots and then create your test cases based on that.
Dinakar
So yeah, I think I think we're good for questions.
Dinakar
There are a lot more questions coming up.
Dinakar
Uh but I think we can take them offline due to the due to a lack of time.
Dinakar
Uh maybe we can wrap it up and uh we can definitely get in touch with you and answer the questions for you if that's okay
Jessica
Sure.
Jessica
Sure.
Jessica
Then Kar.
Jessica
All right.
Jessica
You can reach out to me at any point, you know, and if you are want to discuss more, uh, feel free to reach out to us and we can have a separate demo discussion as well as any other queries you may have.
Dinakar
Okay.
Dinakar
All right.
Dinakar
Great.
Dinakar
Uh so just a couple more slides as to what's the business impact of running gen right.
Dinakar
Uh Jessica would you like me to share the screen again?
Jessica
Yeah.
Jessica
Yeah.
Dinakar
Yeah.
Jessica
So to conclude uh just wanted to take you through how this will actually help be it in terms of uh you know the business impact and the long-term impact for uh QA teams.
Jessica
Yeah, I can see it now.
Jessica
Yeah.
Jessica
So, what really changes when you know our QGEN becomes part of your day-to-day testing workflow, right?
Jessica
And let's look at it from different persona uh aspects.
Jessica
Uh right, so firstly from a QA manager's perspective, uh the most visible um impact would be you know effort reduction, right?
Jessica
So large portion of Q's time goes into coordinating manual test design right and reviewing test cases fixing gaps.
Jessica
Um so with QGNE test cases are generated automatically right and then secondly systemic coverage right you will be able to have a systematic way in which you're generating test cases and that will improve your coverage um and standardization scenario uh based standardization which is actually one of the biggest challenges that we see is actually uh teams having inconsistency across different uh uh you know between different individuals or teams right because different tests testers would write scenarios uh differently.
Jessica
And finally, backlog compression, right?
Jessica
So your test design backlog shrinks significantly.
Jessica
And instead of waiting for test cases to be manually written for every new story, right, QGEN helps you uh generate them as soon as the requirements are available.
Jessica
And secondly, from an enterprise standpoint, right?
Jessica
Uh what are the benefits you would see from an org level, right?
Jessica
you would see faster spin cycles.
Jessica
Uh test case generation would stop becoming a bottleneck.
Jessica
Uh teams move faster from uh requirement to execution.
Jessica
And so this will directly shorten the you know the overall testing prep phase and improve the sprint Uh you will also have traceability where um you know because QGEN works directly from structured inputs and requirements.
Jessica
So you can clearly trace them uh and hence it'll be more audit ready and also knowledge uh independence right so because enterprises often uh depend heavily on maybe a few senior testers who understand complex systems right so with QGEN that knowledge is um embedded and standardized across uh the workflows and finally from a digital you know business perspective right you will have compressed coverage time uh your regressions will become more scalable uh right and um your um risks will be become more measurable because uh you have a systemic way in which you're uh generating the test cases and finally a speed advantage right so you will be able to ship faster um not just by cutting quality but by removing uh manual and repetitive test design uh work right so in summary QGEN will become part of your workflow and uh so it will not just be a tool that generates ates test cases.
Jessica
It will be a way that standardizes quality thinking across uh your QA team and uh the next slide uh yeah what you will be able to achieve is coverage at scale right so uh QGEN will expand your requirements into comprehensive scenarios uh right and you will be able to achieve so that coverage um bottleneck that we discussed uh earlier on right so that would be solved and you would have uh your happy parts, your edge cases, um uh right and you'll be able to test all your different uh scenarios um at scale.
Jessica
Uh effort reduction uh we already discussed that the hours spent in drafting your test cases um and standardization having a standard way of generating test cases so that there is no individual uh dependency and uh time to ship and compression of your uh cycles.
Jessica
Right?
Jessica
So it will um from the time of requirement to ready to run test cases uh you will be able to uh scale up in terms of your speed.
Jessica
So that is the uh you know the organizational level impact you would be able to see uh from a QA uh perspective.
Jessica
So I hope this really gave you uh you know um a new way of seeing how you can generate your test cases, a standardized way of generating test cases and if you have um if this is useful for you and your team um you know discuss with your teams uh or with your managers right and u you can reach out to us anytime uh for a more detailed u customized demo session right maybe even for your application if you're willing to share generate a sample um set of test cases for you and show you how it will actually create that uh impact for you.
Jessica
So yeah, happy to continue the discussions further.
Dinakar
All right, great.
Dinakar
Uh thank you so much.
Dinakar
Uh thank you Jessica for sharing your you know your the wisdom in how to use AI agents for generating test cases and you you've shown us that entire workflow as well and you've added value as to where uh the where is the business impact and how it's being done and that that entire workflow that you shared was very helpful.
Dinakar
Uh thank you for the deep dive uh into this entire topic and so thank you.
Dinakar
Uh we also want to thank each one of our participants uh active participants who joined in uh from the start and we're we apologize for the earlier uh technical glitches uh but I'm sure that some of u some of the ideas that were shared have sparked a few ideas in your minds and to think through and implement in your own scenarios as well.
Dinakar
Uh my recommendation would be to you know mull over these ideas uh talk to us to discuss you there's a QR code right on your screen uh you can just use your phones to you know just reach out to us on our website or go to and you know hit on the talk to us button or start a free trial button and you you you'll be right with us.
Dinakar
So m over these ideas, discuss with us, keep the conversation going and also reach out to us with any questions that you may have.
Dinakar
Uh you can reach out to your by writing to us at info@pcloudy.com and we'll respond to your queries as soon as possible.
Dinakar
Uh with that uh thank you once again for being an engaging audience and happy testing.
Jessica
Thank you.
