Jeroline
Alright.
Jeroline
Uh hello everyone.
Jeroline
Thank you for joining us today from different parts of the world.
Jeroline
Welcome to Pcloudy's webinar on your app is your bank practical AI testing strategies for flawless digital experiences.
Jeroline
We're thrilled to have you with us today.
Jeroline
I'm Jeroline and I'll be your host for today's webinar.
Jeroline
So before we dive in, let's cover a few housekeeping details.
Jeroline
Your lines are currently muted to ensure a smooth experience.
Jeroline
However, we encourage you to ask questions so you can submit them using the Q&A option that you see at the bottom of your screen and we'll address them during the Q&A session that at the end of the webinar.
Jeroline
And also note that this webinar is being recorded and we'll be sharing the recording link after the webinar.
Jeroline
So feel free to share the link with your colleagues or peers who might find this webinar valuable.
Jeroline
And now let me introduce our speakers for today, George Ukkuru and Jessica Selvan.
Jeroline
First up, we have George who's an expert technologist with over 24 years of experience across test engineering, product management and user experience.
Jeroline
He currently serves as the general manager at McLaren Strategic Solutions where he drives cutting-edge initiatives in software testing, automation and AI-driven quality assurance.
Jeroline
George is widely recognized as a thought leader in the testing and automation space.
Jeroline
He's a LinkedIn Top Voice, a frequent speaker at global industry events and the host of the popular podcast Automation Hangout.
Jeroline
Over the years, he's helped numerous organizations across industries enhance their QA maturity, accelerate release cycles and implement scalable AI-powered testing solutions that drive real business outcomes.
Jeroline
And we also have Jessica who is a pre-sales specialist at Pcloudy with over five years of experience in the IT industry.
Jeroline
She's specializes in test automation, AI-driven testing strategies, and continuous testing methodologies.
Jeroline
With her expertise in modern testing frameworks, DevOps integration, and AI-powered test optimization, she's helped numerous organizations optimize their testing processes.
Jeroline
So, without any further ado, uh let's welcome George and Jessica.
Jeroline
And George, over to you to take this webinar forward.
George
Uh thank you, Zaroulim.
George
And uh good morning and good evening to everyone.
George
So, we're primarily going to look at uh I would say a couple of use cases around uh specifically uh focusing on the digital bank.
George
So, we'll actually get started by looking at some of the interesting trends that we have seen in the digital banking side.
George
So, you look Look if you look at the overall banking industry in the last couple of years, there has been a significant shift where customers are more comfortable using digital banking channels compared with maybe uh any branch kind of a visit.
George
And I would say 3/4 of the customers that any bank is having are actually using digital banking currently.
George
And due to this, there's a lot of focus uh in the on the banking side uh to actually bring in innovation as well as transformation uh to the digital banking channels so that you provide a better experience as well as generate additional revenue and So, let's take a look at some of the maybe the top seven trends around uh digital banking.
George
So, there has been a huge, I would say, adoption of Agile across industries, and banking is not, I would say, different.
George
So, most of the banks have decided to go with the Agile model of development.
George
And you're primarily having either a three- or four-week sprint in which products are actually getting released into uh production.
George
And this is actually helping banks uh to actually um meet the uh regulatory compliance as well as they're also able to provide additional features or new features to their uh customers.
George
But, from a testing side, this actually puts in a lot of pressure on testers uh because this is going to you're actually going to get very less time for your test design, test planning, and test execution Now, there's a lot of focus around, I would say, hyper-personalization, which means that you're looking at the customer usage patterns, behaviors, age groups, lifestyle, purchase patterns, and all of those things.
George
And you're primarily using AI and analytics to actually push customized offers to customers.
George
So, customized offers can be in the form of a home loan, credit card, or it can be a personal loan, depending on the analytics and the AI output.
George
Now, in today's world in digital banking, you actually have numerous channels through which you can actually bank.
George
ATMs, you have kiosks, you basically have internet banking as well as mobile banking.
George
And it's very important to actually provide a unified experience across all these channels.
George
And also important to make sure that you to look at scenarios like where the customer starts a transaction maybe in internet banking application, and then maybe ends it maybe on a mobile app.
George
So, there should be a smooth transition between channels as well as the unified experience across channels is also very from a banking standpoint.
George
There are a lot of frauds and a lot of hacking and other things happening on banking websites.
George
Due to it, there's a heavy focus around security.
George
So, the current policy of bank is something like never trust, always verify.
George
That means like they are very cautious in actually onboarding customers.
George
They also do a lot of verifications for every transaction that you do.
George
So, security is also having a very big impact around digital banking.
George
And banks are also heavily investing on the security and other other related activities.
George
The cloud has actually been considered as I would say, banks have become open towards cloud adoption.
George
You can actually see that many of the banking solutions are now moving into the cloud.
George
And primary reason is to achieve more scalability as well as reliability.
George
And we have there's also focus around continuous monitoring, which is nothing but looking at what your users are doing in production.
George
It can be I would say monitoring the performance of an application, monitoring health of an application, or even monitoring customer feedback in real time.
George
And this data is continuously used to proactively solve issues as well as define new strategies for what I would say rolling out Last but not the least is around AI-driven development or AI-driven solutions.
George
You can actually see a numerous amount of solutions that are AI-based being used inside a bank.
George
Starting from maybe a chatbot or a virtual assistant with whom you may be what I would say doing conversations to solve your tickets.
George
Or it can be an AI model that is being used by banks maybe to issue loans or maybe to detect what I would say fraud-related transactions.
George
So there are numerous, I would say like a lot of investments that have gone into in banks to actually implement AI.
George
And you you also you can also see that you can also see that this is actually going to increase in the next couple of years as well.
George
So these are some of the common trends that you see among digital within the digital banks across the globe.
George
But the expectations and the and and and the the expectations are very high and there's a lot of transformation happening in the digital banking side.
George
But at the same time, when you look from a software testing angle, there are a lot of challenges that tester or anyone working in a testing group.
George
It can be a test manager or a test architect.
George
There are day-to-day challenges that we may have to deal with, especially when it comes to banking.
George
One is basically there is there is a high pressure on banks to actually comply with regulatory requirements.
George
So that's the reason why they have adopted Agile.
George
So it's almost important to actually make it's a very important from a testing standpoint to ensure that the releases are actually happening on time.
George
But many times the tester testers or the testing team is slow in providing feedback to the development team.
George
And what happens over a period of time is like you may actually miss on your deadlines or your release schedules because you are actually giving the defects or reporting defects maybe in the last week of the sprint, and the developers may not get enough time to solve them, and finally your release may not happen on time.
George
Discovery then quality seems continues to be a challenge especially with highly critical applications or modules or capabilities like payments as well as I would say like money transfer.
George
So, the coverage is slow and the quality is also sometimes very concerning with digital channels.
George
Since you actually have releases happening every week or every month, the amount of changes that happens to the user interface is very high, and due to which even if you have a good automation coverage, you may actually end up with a lot of test case failures when you're trying to run these regression packs as part of your regression cycles.
George
There's a lot of data that is being collected both from a software engineering standpoint as well as from the customer, but very rarely I've seen we have seen testing teams utilizing this data to improve the way testing is done.
George
So, it can be maybe production defects, or it can be production tickets, or maybe feedback that is actually coming up in social media forums like Twitter.
George
Testers do not pay a lot of attention to this kind of feedback, and it is very rarely used to improve the way in which testing is currently being performed.
George
Even though applications have moved to the cloud or are moving to the cloud, performance seems to be a concern.
George
There has been improvements in the area around performance.
George
You rarely see banking sites coming down, but you can notice that there are there are time periods where you actually see slowness in both your mobile banking as well as with your internet banking applications.
George
So, security seems to be So, sorry, the performance seems to be a concern, and there's not enough attention that is being paid to performance as well as non-functional testing in many of the Since what I would say like you're working in a design mode, hardly get I would say a couple of days or a couple of hours to actually complete your test design.
George
Due to which you may actually end up writing low-quality test cases.
George
And moreover, you may be under heavy pressure to complete your test cases before the first building is given to the testing team.
George
There is no way in which you can actually find out how good is your coverage unless you're actually using maybe a sophisticated tool or an AI-based tool.
George
So in many organizations, the regression coverage as well as the overall functional coverage is not known because it is very difficult to find out what amount of coverage is being offered by the test cases that are that are already there.
George
And last point is around inclusiveness.
George
There are there are there are a good amount of I would say folks who are differently abled who are digital banking users, but there I I don't think even now we have enough focus around this area of accessibility.
George
So accessi- accessibility and inclusiveness seems to be a concern with banks and this is seriously impacting their revenue as well as the user satisfaction.
George
So these are some of the challenges that you see uh in the in with which are related to software testing within banks.
George
Some of them may be across industries, but there are maybe very specific challenges like quality as well as performance which are highly which may highly impact the performance of a digital bank.
George
Now let's likely look at how is that we uh use artificial intelligence or how does artificial intelligence help uh a software tester.
George
One is definitely productivity.
George
Uh software testers can be more productive uh if if uh by using AI-based solutions.
George
The productivity improvements then can come in in terms of test case design, test case execution, or test We'll we'll see that in detail on how it actually provides you productivity improvements in the subsequent slides.
George
Similar to productivity improvements, the next area where testers can I can see benefit with AI is improving the speed of testing.
George
So, improving the speed of testing is primarily by maybe you can actually cut short your test execution cycle.
George
Maybe by using AI, you can actually select the test cases that are of high importance and execute them.
George
Or you may actually use AI to actually shortlist test cases that have the highest amount of highest probability of finding defects.
George
So, AI can definitely help you to save a lot of time in across various activities in the testing life You can also use AI uh to actually take decisions.
George
As I was mentioning earlier, there a lot of there's a wealth of data that is actually collected both from the customer as well as from the uh what I would say from the software engineering standpoint, which is very rarely used to generate insights.
George
But, you can actually use predictive analytics and other solutions AI based solutions in order to generate insights, which can actually help you take decisions.
George
And AI can also help you to actually shift left your testing.
George
There are many ways in which you can actually carry out testing much earlier in the life cycle with the help of AI.
George
And this also can actually help you to reduce the overall cost associated with the rework.
George
A simple example can be maybe finding out maybe regression defects early in the life cycle and maybe by doing a targeted regression based on coding fact.
George
That can be a good solution that can actually help you to find regression defects maybe in the first week of your sprint.
George
There's always a focus on user experience and AI also can actually help you to improve the experience.
George
Visual testing or maybe verifying the visual components of an application is a good example where where you can actually improve the overall experience that is provided to the customer.
George
Or it can be an AI based solution that can be used to analyze customer feedback that that which can be used to improve the way testing is done.
George
So, there are numerous ways in which AI can be used in order to improve the customer experience.
George
Finally, you can also look at the operational overheads.
George
There are a lot of there are a lot of companies where we have seen test reports being prepared manually.
George
There's a lot of dependency on maybe infrastructure team to provision data.
George
So, a lot of self-service kind of solutions can be implemented like a chatbot which can actually provide you test reports or maybe insights around what's happening with your release.
George
Those kind of solution which are self-service can also reduce a lot of manual activities or overheads that are associated with reporting as well as some of these can be related administrative type of work as well.
George
So, these are some of the way these these are some of the what I would say benefits that a tester would actually get or a testing team would actually get by implementing an AI-based solution.
George
And definitely, I would say in the last 3 years the maturity of AI solutions have increased.
George
Previously, we were actually creating custom models and uh other solutions to implement anything around AI.
George
With OpenAI, Deep Search, and other models being in place, there are a lot of products and I would say solutions that have come into the market that actually have AI capabilities.
George
And the maturity of these solutions have also increased significantly in the last couple of years.
George
So, it's very important to look at AI very seriously now.
George
Even though it has not reached a state where it can take away testing jobs, definitely, it is actually uh uh I would say organizations uh who have implemented AI, they're really seeing the benefits.
George
It can be I would say in terms of cost savings, effort savings, or in terms of quality.
George
Now, let's take a look at some of the common use cases or some of the top use cases around AI.
George
You can actually use AI to actually analyze user stories.
George
You can actually use AI to generate test cases.
George
You can actually do an impact analysis to find out areas of your application that needs to be tested.
George
You can use AI to actually reduce your test maintenance overhead by implementing self-healing.
George
You can use uh visual testing to to verify the visual components of your application as well as to make sure that your application behavior is consistent across different devices, browsers, as well as resolutions.
George
You can use coding assistance, which is something new that came into the market a couple of years back.
George
Coding assistance is another AI solution that you can use in order to improve your productivity around test automation.
George
You also have test observability, which is mostly, I would say, shift right testing technique, which actually helps you to ensure that your application is continuously available in production.
George
And then you can also use AI for test orchestration, primarily to execute a large number of test cases in a minimal amount of time.
George
So, these are some of the use cases.
George
If you look at the complete list, I would say there can be around 30 to 40 different use cases that you can think about from an AI standpoint.
George
And I would say like if you're actually planning to implement AI in your organization, you can actually look at your current challenges and then maybe decide what kind of solutions may actually help.
George
You're just giving you a sample, but I would say a very short list of use cases.
George
In the interest of time, what you've done is you've actually shortlisted around five use cases from what we So, the first use case that you would like to take up is basically smart user story analysis.
George
So, requirements or user stories.
George
The requirements no longer exist because every organization moved to Agile.
George
So, primarily user stories is what you see in every project.
George
And very rarely I've seen the testers spending enough time in order to groom the user story or to analyze the user story to find anomalies.
George
So, what typically happens is the first week of the sprint or before the sprint grooming sprint grooming session or planning session happens, testers don't open up open openly tool and look at the user stories.
George
The primary focus is just to do planning and some kind of estimate without even understanding the user story.
George
And maybe the first week of the sprint when you're supposed to write test cases, you go through your user stories and then find out there is not enough clarity around the user stories and then you start raising clarifications with your product owner or with your business analyst.
George
And you lose a lot of time.
George
Your test cases may be half cooked or you might have written your test cases already.
George
Your developers would have already started coding.
George
They might be doing unit testing.
George
At that point of time, you start your business analyst or product owner may start making changes to the requirements.
George
And then you actually end up with a lot of rework.
George
Or in some cases, you may not even know whether the user story was good enough.
George
When you once you implement the product, once it goes to production, based on customer feedback, you may find out that the requirement that your business analyst has given you were incomplete.
George
But now, the only there are multiple ways in which you can solve the problem.
George
You can actually focus You can actually spend I would say a good amount of time analyzing the user stories manually before your grooming session.
George
Or you can actually have what I would say spend time reviewing and discussing about the user story during your grooming.
George
Which is actually, I would say, a manual activity which which consumes time.
George
And if you want to really participate in those kind of meetings or activities, you also need to have some kind of application as well as domain expertise.
George
Which may not be available always, especially if you're working in the services sector.
George
But now there are which can be either custom or commercial solutions that you can leverage.
George
That can actually take user stories from a tool like Azure DevOps or Jira and then start analyzing those user stories using any kind of industry framework.
George
There are a lot of frameworks around user stories that are very popular.
George
SMART, INVEST, QUEST are some examples.
George
But there are many more user story frameworks that are available.
George
You can actually use an AI-based solution to review user stories and then find out anomalies uh in those user stories.
George
Then you can also go one step further.
George
You can also train these solutions with your project requirements or with your application documentation.
George
So that it actually gives you very meaningful, I would say, results.
George
And what you can do, you can actually go ahead and once you actually have these what I would say anomalies, ambiguities, I would say, listed, you can either go and fix it manually with the help of your business analyst or product owner.
George
Or you can actually use the tool itself to actually rewrite your user story as well as acceptance criteria.
George
So if you look at the data, even now there are like more than 70% of the projects that are actually failing because the requirements were not captured in the right manner.
George
And you can actually get rid of this problem to a very great extent by actually using user story, what I would say an AI-based user story analysis, which is something that you can easily do.
George
Maybe after maybe if you spend a couple of weeks or maybe like two to three months of time, you can even develop your own solution based on your based on your company standards as well as based on maybe some kind of model training that you can actually do based on your application documentation as well as domain.
George
That is first use case.
George
I've actually I've started using this I started using this I would say maybe three years back.
George
We developed a custom solution and then when Open AI came in, we started implementing this as using the APIs of Open AI and we've actually seen good results where we're able to detect up to 90% of the issues with respect to user before you actually start the sprint.
George
And then that has actually helped us to achieve a good amount of sprint velocity as well as the commitments that we provide to the business team.
George
The second solution is around test case generation, which is an area which is I would say maybe the second highest in terms of effort.
George
Test execution of the I would say is is the phase or activity that takes the highest amount of time in testing.
George
Ideally, you spend 30 to 40% of your time in test execution, whereas you may spend around 20 to 25% of your effort in designing test cases.
George
You definitely the manual approach is you you review a user story, you analyze it, you basically break it down into test scenarios as well as test conditions and then you start manually writing these test cases.
George
And if I give a user story to two testers, you may actually get maybe different number of test cases and different number of And it is primarily I would say experience-based.
George
Even though you use standard techniques, there can be differences in the quality of test cases that are actually generated by testers and experience plays a very major role when it comes to uh case design.
George
Now, and it's actually effort intensive as I was mentioning earlier, 25% of the effort is actually being spent on test design.
George
But, now there are a lot of ways in which you can actually automate the uh test design.
George
So, prior to maybe AI's uh becoming popular, you might have heard about model-based testing tools, Agile Designer uh was there, there are a couple of other tools that were available in the market that actually had this capability of maybe taking workflow diagrams and then generating test cases, and they were very good from an end-to-end testing uh standpoint.
George
But, now there are multiple other inputs that you can actually take in order to generate uh your test cases, and in some cases there are solutions that can also generate automation scripts.
George
One is basically, if you're looking at maybe an application that is built using microservices architecture, you can actually look at, I would say, API as their input.
George
You can actually refer to a Swagger documentation, and AI-based solutions can actually generate test cases from your uh API documentation.
George
So, these are very fast uh API tests are very fast, they're less uh costly to maintain.
George
So, API should be can be a first target when it comes to test case generation, or you might be working on an application that's already there, or you might be having a prototype but no requirements.
George
You can definitely come up with a solution or design a solution that can actually look at maybe a Figma prototype and then generate generate test cases, or you may have user stories that are actually groomed.
George
You can actually feed in user stories to your test design tool, and the tool can actually give you a set of functional automation scripts or functional test cases, depending on what you need.
George
And this also gives you what I would say, like this also helps you to what I would say achieve a lot of coverage, or the optimal amount of coverage, as well as it also saves a lot of time.
George
Or you may actually use a solution which is primarily, I would say, maybe a prompt-based solution where you actually type your requirements maybe in simple English, and the tool will actually generate an automation script or a test case based on the instructions that you provide to the tool.
George
So, there are like five or six four or five different ways in which you can generate test cases, but most of these tools will actually have an NLP parser, a test case generation engine, a plugin that actually helps you to connect to a user management tool, a user story management tool, as well as a formatter which actually helps you to generate test cases in the required format.
George
So this is I would say a beautiful solution that's out there.
George
There are plenty of tools that can actually help you.
George
Worst case, what you can do is you can actually create a custom GPT with a prompt which can actually help you to automate this process around uh test case generation.
George
And and And from my experience, the quality of test cases that you get from these tools is dependent on the prompt that you write, as well as the engine, if it's a commercial tool.
George
But the some of the tools that I saw, the quality of output was actually equivalent to the output that you'd actually get from from a tester who's having maybe two to three years of […]
George
The next important area where uh uh AI can actually help is basically self-healing and test orchestration.
George
I've seen organizations where there are like thousands of test cases that needs to be run.
George
I I used to work with a customer, a technology customer, who actually had around 34,000 test cases or test automation scripts that they were running on their global application.
George
And it used to take days together to run this.
George
Now, you actually have solutions which can actually help you to do large-scale execution.
George
Most of the automation tools that you see in the market actually has a parallel execution engine that can actually help you to run multiple tests in parallel.
George
You can actually configure uh the engine in such a way that you can decide upon the number of tests that you want to run in parallel.
George
And then most of these platforms actually have an intelligent orchestration engine that decides upon where you should be running your tests on based on what trigger.
George
So that's what an orchestration engine would primarily do.
George
You may decide to run your tests maybe on a maybe a mobile device platform or a cross-browser platform.
George
Or you may you may have your own infrastructure in the cloud with Docker's where you may be running your tests.
George
But you have to actually use your resources in the right way.
George
So, it's very important to actually make sure that you're actually completing your test in the shortest amount of time using the least amount of resources.
George
So, now there are AI-based solutions which actually has intelligence to actually what I would say distribute your workload across your what I would say environment and then complete the test execution at a much faster pace.
George
So, this actually helps you to I would say complete your regression cycles or your functional test cycles in a very short amount of time.
George
Now, even if you have a very good parallel uh execution engine as well as an orchestration engine, most of the automation projects may not give you the required results primarily because you end up what I would say running your scripts and then maybe like 20% or 30% of your scripts will not will not be actually functional when you're running it for a formal regression cycle.
George
The primary one of the primary reasons due to which uh the the scripts won't work as maybe because of the changes that has actually happened to the user interface.
George
Maybe the developer has actually fixed the bug and they've changed the label.
George
Or you may actually have a scenario where there's a new control that has been introduced.
George
Or it can be a scenario where the properties of an object is actually modified.
George
Or maybe the control itself was actually So, there are a lot of scenarios a lot of what I would say 80% of the time the test scripts actually fail due to I would say like script-related error.
George
There can be also scenarios where your automation script might fail because of synchronization issues.
George
So, this may be due to maybe application behaving too slow or too fast or either of these.
George
Or it can be an environment-related issue which is very sporadic.
George
It or maybe for a second or so your environment was not accessible.
George
So, there are like two to three ways in which you can actually solve these problems.
George
So, these are the three problems and you can actually solve all of these using a the self-healing engine.
George
For example, if you're not able to find the property of an object, there are there are alternate locators that you can use.
George
There are multiple ways in which you can recognize an object.
George
Maybe XPath is one of the method.
George
You actually have class, ID, and other mechanisms using which you can actually learn the properties of an object.
George
So, what you can do is So, what the AI engine would actually do is it actually looks at alternate property that it can use in order to identify the project identify the object.
George
It rescans your UI, learns the object, and then try to execute this code.
George
And if you're actually facing a synchronization issue, the tool will actually have the intelligence or the solution actually have intelligence to actually go and modify the wait time so that the script actually is synchronized with your application.
George
Or if it can it can be an environment not being available, you actually do have any retry engine that will actually go and find out what's happening with your environment by actually maybe waiting for some time and then triggering the automation script.
George
So, self-healing is actually a very beautiful tool.
George
You can see up to 80% of test automation maintenance cost getting what I would say it reduces your automation maintenance cost at least by 80%.
George
You can develop your own solution if you're actually having an open-source framework or platforms like Pcloudy actually has this kind of feature now, which can be used to what I would say auto-rectify your scripts with no effort.
George
So, what typically happens is you may actually have maybe 1,000 test cases, 20% of scripts fail.
George
Your automation team will not have time to fix them.
George
So, what what happens is you start executing these 20% of scripts manually.
George
You may not be able to achieve your deadline.
George
Your testers will actually get stressed out because they have to complete a large number of test cases that have failed.
George
And now, after your test execution is completed manually, you have to go back and again fix your script.
George
So, it's actually very what I would say there's something that happens in every release or every test cycle, and you can actually avoid this pain by actually looking at implementing something like self-healing.
George
Now, moving on, the next use case is around coding assistance.
George
You might have heard about coding assistance being used by developers.
George
GitHub, Codium are some of the coding assistants that are available.
George
You also have Amazon Whisperer.
George
There are a number of coding assistants that are available, which are primarily, I would say, tools or enablers that likely make a tester or a developer more productive.
George
From a test automation standpoint, you can actually write very simple prompts that can actually help you to generate automation scripts.
George
It can be like you providing a URL and then explaining what links to click and what actions to perform in form of a very simple English text.
George
And these these prompts can actually help you generate scripts.
George
Or it can be an API definition that you can actually provide as an input to these I would say coding assistants and it can actually take those APIs and then design a set of test cases based on the So, test script development can be actually I would say accelerated by using solutions like uh coding assistants.
George
It is not just test script development, you can actually use uh a coding assistant to maybe even understand the code that has been written.
George
For example, you've been given with a framework maybe for a tool like Playwright and you want to understand how the framework is written.
George
You can actually use a coding assistant to explain the code.
George
Or if you're very new to test automation, a coding assistant can actually help you uh learn uh the test automation tool as well as the programming uh what I would say language associated with it.
George
Most of the cases coding assistants are used with uh I would say open source frameworks.
George
They're those are They are primarily used for I would say scripted kind of framework, but now uh there are codeless platforms also that actually has a coding assistant which actually helps testers to generate test cases as well as fix issues in a much faster mode.
George
There are other use use cases for coding assistants as well.
George
They can actually help you to identify the runtime errors that are caused.
George
For example, you create a script and it's actually giving some kind of error due to a programming error or maybe due to an object recognition issue.
George
You can actually put those errors into the coding assistant and the coding assistant will actually help you to fix the issue.
George
Or you've written a large amount of code as part of your framework and you want to refactor as well as document it, coding assistant can be a very good choice.
George
So, there are multiple use cases uh for coding assistant.
George
On an average, if you're an automation tester, and if you're actually uh involved in script development and maintenance, a coding assistant can actually save you at least 1 hour of work every day, which is almost equivalent to maybe, I would say, 20 to 22 hours of work every month.
George
So, that means like half of the week's work, you're actually working half a week extra, or you're actually being more productive by using a coding assistant.
George
And these are not very costly.
George
Some of the coding assistants comes free of cost, especially GitHub Copilot, and they're actually available for what I would say, for a very nominal amount, or you can actually also use a free version.
George
They're very productive, I would say, very useful tools, especially for engineers who are actually working Playwright, Selenium, or as well as other kind of tools like Cypress.
George
let's actually take a look at the last use case, which is around test observability.
George
You might have heard about shift left, which is actually what I would say, bringing testing to the what I would say, to the beginning of your test cycle or agile sprint.
George
For example, the first use case that we saw, um the user story analysis of the smart user story analysis is a good example of a shift left testing technique.
George
Testing your APIs, as well as continuous regression testing, are good examples of shift left.
George
Now, test observability is about shift right.
George
Now, many times what happens is your testers, or the testers working in a bank, or maybe in a financial service organization, they actually pay very little attention to what happens to your application after it has actually gone live.
George
So, there may be production tickets that are actually coming in.
George
You The The end customers may be actually experiencing performance issues.
George
The customers may not be happy with your application or the features that you've delivered in the last release.
George
They may be actually posting nasty comments in social media.
George
You rarely spend time in analyzing any of those.
George
But, there are a lot of ways in which you can actually improve the experience that you're providing to customers.
George
And test observability, I would say, is I would say, one of the good practices or the one of the best practices that they can implement from a ship right standpoint where they're actually doing testing or monitoring on a production instance.
George
So, you can actually do maybe synthetic user monitoring as well as real user monitoring are two techniques that you can use or approaches that you can use for monitoring applications.
George
So, your monitoring is to make sure that your application is functioning correctly in production.
George
And it is not just functional aspects, you can also see the response time as well as the performance of your application in production.
George
So, that's the second use case.
George
So, like a monitoring what I would say solution can actually help you to actually simulate transactions.
George
This is this basically called synthetic user transactions.
George
For example, if I'm working with a digital banking application, I may be using maybe I may be actually performing transactions maybe like checking account balance.
George
I may be actually logging in.
George
I may be making sure that the loan application page is working.
George
So, I may be defining transaction maybe by reusing some of the regression test packs that I may be having.
George
And I may actually run it on the production environment to make sure that your application is functioning properly.
George
So, you may actually run scripts maybe every 1 hour and then see whether your workflows or your test scripts are actually passing and nothing is broken in production.
George
At the same time, you're also monitor the performance of the application.
George
You may also see okay, how much time did it take to complete a transaction like verifying account balance?
George
Is it within the acceptable limits?
George
Or is my application actually taking are my API calls taking a lot of time?
George
So, all of these can be actually verified using synthetic user monitoring.
George
So, there's no real user here.
George
You're actually using an automation script to simulate actions and that's what the basis behind synthetic user monitoring.
George
Now, when you're doing synthetic monitoring in the production or synthetic user testing in production, parallel to this, your real users also might be using your So, they might be performing real transactions.
George
So, you can also monitor those transactions And the objective here is to basically look at again the performance as well as the application health.
George
And you may also find some kind of symptoms or anomalies in some of the transactions, which can be a good clue for your support team or your SRE team to go and fix the issues before it is actually getting reported by a larger user base.
George
So, real users will be actually using your application.
George
You'll keep monitoring them.
George
As well as you'll also have synthetic users or scripts being run parallel to this.
George
It will also help you to measure.
George
Maybe the amount the type of metrics that you monitor may be very different.
George
For example, when you're doing synthetic user monitoring, you may be looking at I would say things like a page load time, API response time, app availability.
George
Whereas when you're looking at real user monitoring, you might be interested in seeing error rates, throughput, maybe last contextual pain and things like that.
George
So, there is overlap between the metrics, but there are certain set of metrics that you specify for your RUM versus this tool.
George
It's a very useful technique.
George
The testing team also should get involved in this, especially around synthetic user monitoring, where they can actually reuse their automation scripts in production.
George
There are certain precautions that you may need to make.
George
You may actually have to use maybe certain customers or certain set of data to test in production.
George
Or in some cases, you may have to only run scripts which are actually read only.
George
So, the other important aspect which I forgot to mention around user monitoring is you may be able to simulate workloads from different parts of the world.
George
So, depending on where your application is hosted, you may be able to track your application performance.
George
So, that's another way in which it's not just making sure the application is available.
George
You can also see the response time across different countries when people are logging in.
George
What would be the response time based on where from where you're running your scripts.
George
So, and in most of the cases, you may be running your automation scripts during synthetic user tests in a cloud environment or maybe in a platform like Pcloudy.
George
So, this is the fifth use case.
George
So, since you've covered the major I would say the top five use cases, I'll hand it over to Jessica now, who's actually going to give us an overview of the AI capabilities of the Pcloudy platform.
George
And she's also going to give us a couple of demos on some of the AI capabilities within the Pcloudy platform.
George
Over to you, Jessica.
Jessica
Thank you.
Jessica
Thank you so much, George, and for sharing your insights based on your vast experience and particularly insights in the you know, the testing and the in the banking space.
Jessica
So, firstly to introduce what Pcloudy is, right?
Jessica
We are an AI-powered digital experience testing platform and we the core of our platform is our diverse device coverage.
Jessica
So, we have a 5,000 plus device browser combinations, real devices that are hosted in our data across globally in multiple locations.
Jessica
we also do support custom devices and we also have secure deployment options, which is also something that is that flexibility is required particularly in the banking space with on-premise lab in a box which is a quick to implement on-premise setup along with a secure private cloud, which would be a dedicated tenant with dedicated devices for you 24/7, as well as a shareable public cloud.
Jessica
And the first layer of solution that you see here is the functional experience testing and you'll be able to manually as well as through automation access these our device infrastructure.
Jessica
You can manually go ahead and do your functional validations, run your we support integrations with pretty much all the key frameworks, right?
Jessica
Like APN, Selenium, Espresso, Playwright and so on.
Jessica
And we also have specialized features that are particularly relevant in the banking space like you know, OCR, image injection for your check scanning or your ID scanning along with biometric simulations and so on.
Jessica
And next is the performance experience testing.
Jessica
So this is more on understanding how your app is performing and how it could potentially be impacting the digital experience of your users, right?
Jessica
Something that also George was mentioning in terms of the criticality of identifying lot of issues that are identified in production.
Jessica
So to preempt that and to ensure that you're running your performance experience you know test from the app level, not just on the load testing side, but in the app level tracking all the key metrics like your UI performance, what are the page transition times, what is the time taken for your app to what are the backend network API calls that are going on, are there any specific API calls that are leading to some delays or impacting the performance smoothness of your application.
Jessica
Even on the UI side like jerky frames and of course things like you know your CPU spikes, your memory, your battery consumption, your network data and so on as well as analysis of logs.
Jessica
So all of this that could be potentially impacting the smoothness of your application, the experience of your application 60 plus metrics.
Jessica
So this is a key differentiator that we've been able to you know um bring about here at Pcloudy.
Jessica
And finally on the digital experience monitoring or synthetic monitoring as well that George was referring to where in your production application as well you'll be able to run critical workflows 24/7 and generate these similar insights and get those anomalies and performance related insights.
Jessica
So the digital experience and synthetic monitoring is also a space that we are venturing into here at Pcloudy.
Jessica
And also the AI agents now I want to just take a minute to talk about the AI agents that are driving this you know this innovation here.
Jessica
Firstly you know in the build part of it we have a test creation agent which is Qpilot another key differentiator at Pcloudy where we've been able to um um have a fully autonomous end-to-end agent uh that can create your test script uh and generate it.
Jessica
Uh you'll be able to club it as collections and run your suites across cross-platform omnichannel again that is useful for in the banking space, you know, you be it your web applications, your Android, iOS, irrespective of you know, the technology that your app is built on, right?
Jessica
As well as on the execution front self-healing as well.
Jessica
So, um being able to fix uh the flakiness of your scripts, improving test maintenance.
Jessica
Uh the orchestration agent as well, uh right?
Jessica
Being able to smartly allocate uh devices to optimize the time and your resources uh on which you're executing your automation.
Jessica
And finally on the observability and impact analysis side, so this is more from tracking your performance of your application, um how is your uh the AI ML-based insights across different aspects to see how your application is performing.
Jessica
So, this is uh to just give you a quick overview about the entire solution in our platform.
Jessica
The solution layers uh right?
Jessica
And um uh us being the forefront in the digital uh testing space.
Jessica
So, with that I'm just going to go ahead and um you know, just give you a quick idea into a few key AI agents.
Jessica
Of course, we don't have the time to get into live demos, but of course, if you're interested, please uh feel free to check out all our other webinars where we've had live uh demos as well as um uh you know, you can reach out to us if you're interested uh to know more.
Jessica
So, with that I'm just going to go ahead and share my screen.
Jessica
So, first we'll be looking into the self-healing feature.
Jessica
So basically what we're doing here, let me just set the context, is that we have applications, a version one of the application based on which I'm developing a script, right?
Jessica
So you can see that I have the version one of the application here and I'm going to run this particular sample script on a Pcloudy device, right?
Jessica
So we're going to first go ahead and inspect this and see how it is you know the object properties of the email, password, and sign in, right?
Jessica
So for example here you can see it's […]
Jessica
Uh right?
Jessica
And similarly we have developed the script accordingly with input_username as the ID.
Jessica
Similarly for password input_password and for login button_login, right?
Jessica
So this is how the you know this is the for the first version of our application, these are the object properties and we have developed our script based on that, right?
Jessica
So now I'm going to go ahead, you know, and execute our script and we're going to select a Pcloudy device and we have selected a Samsung Galaxy S22 and we're running the script.
Jessica
So this is just a sample APM script and you can see in the my active sessions the automation has gotten triggered and we're going to get into the live view.
Jessica
You can see that in this case the test step has passed.
Jessica
So this is expected because it's our first build and based on which we have developed our script and our automation script and the test case has passed.
Jessica
Now we're going to look into a scenario where in cases where you know we have the version two of the application and the object property has changed, right?
Jessica
So we're going to look into the second build, version two of the application as you can see and we're going to go ahead and inspect the object properties of version two of the application.
Jessica
So, we're going to get into the objects by mode at Pcloudy which allows us to you know, get the object properties of your application.
Jessica
We're getting into the objects by mode and here you can see that instead of username it's changed to input underscore user.
Jessica
But we're going to run with the same script that we had.
Jessica
And similarly even password is changed to a different name and similarly for login as well.
Jessica
So, all three object properties have changed, right?
Jessica
And we're we're going to go ahead and run the same script on the version two build of our application.
Jessica
Let's see how that works.
Jessica
We're now going ahead and running that.
Jessica
You can see that the automation has gotten triggered.
Jessica
We're getting into the live view.
Jessica
You can see that the test step has passed exactly like how it was for the first version of the build.
Jessica
So, this as George was mentioning would really save in terms of your test and script maintenance where it's able to identify that there has been an object property and it will scan for the next best object property, right?
Jessica
And still ensure that your test step passes.
Jessica
So, it prevents the flakiness of your test steps and improves your maintenance, drastically improves the time taken for your script maintenance, right?
Jessica
That is the quick overview on the self-healing part.
Jessica
Next, let me go to another feature.
Jessica
So next we're going to be talking about Qpilot.
Jessica
So Qpilot is an agentic AI agent which will help you autonomously generate your test scripts, right?
Jessica
So just to quickly we're getting into we're connecting into a device and getting into the Qpilot mode.
Jessica
We're selecting our application, giving the app package, app activity, right?
Jessica
And I'm selecting uh the feature name.
Jessica
And I'm going to go ahead and give the description in plain English language as you can see, right?
Jessica
There is no you don't there is no standard format required here.
Jessica
You can even add validation steps, uh right?
Jessica
And you can add test data if required if you don't want to input static data.
Jessica
So it is just pretty much you don't need any coding skills.
Jessica
I'm just giving my natural language and I'm going ahead and giving my test steps.
Jessica
Post which you can go ahead and save your test case.
Jessica
And you can go ahead and execute it.
Jessica
So we're Now we're going to go ahead and generate the test script and the agentic AI agents kick in, right?
Jessica
So we have an X path locator where it will identify the unique X paths.
Jessica
You can see each step is getting executed ensuring high accuracy of the script.
Jessica
So you don't need to spend time debugging.
Jessica
That is the key differentiator we've been able to achieve here at Pcloudy with uh the accuracy of the scripts being really high, right?
Jessica
You just simply need to go ahead and trigger your scripts.
Jessica
You don't need to spend that time debugging, right?
Jessica
And you can see that before the generation of each step, it executes and ensures that it's accurately functioning.
Jessica
Post which it will generate the entire script.
Jessica
Yeah, now we're just going through the flow.
Jessica
This is a sample demo banking app in which we have created a recharge workflow.
Jessica
Once the test script is generated, we it will redirect us to the Qpilot Um and yeah, I want to quickly highlight that it works the similar way for your web application as well.
Jessica
You will simply connect to a browser, uh go ahead to the Qpilot mode, give the URL, uh a test description, save your test case, and it will go ahead and generate the script, right?
Jessica
So, this is again a sample website URL that we've taken, and the test steps are getting generated, and it will redirect to the Qpilot dashboard.
Jessica
So, it works pretty much seamlessly both for your iOS, your Android, as well as your web […]
Jessica
Now, we're getting into the all tests in the test collection.
Jessica
And we're getting into our particular script that we've generated.
Jessica
You can see that we have the description here, and this is the fully generated script.
Jessica
So, this is available in multiple languages like JavaScript, Python, Java is also going to be released soon.
Jessica
Right, you can copy and paste this into your own frameworks, as well.
Jessica
Uh we have test data that will be inputted here, and I can simply go ahead and run my test case individually, or I can group it into a collection over here, and run it as a […]
Jessica
So, you can see here that I'm going to add a few cases here.
Jessica
We're adding multiple test cases.
Jessica
We're selecting the version range of devices that we want it to execute on, saving the test case, and clicking on run.
Jessica
So, I'm running this from the Q already UI, but you can also integrate this with your CI/CD and run it from there, as well.
Jessica
So, post your run, you'll be able to see the live view, as well as the reports and the status of each of your test case, whether it's passed or failed.
Jessica
And for each and every test case, you'll have the test logs, the device logs, your APM logs, as well as the app profiling with CPU and memory details, with the video of the entire session, right?
Jessica
So, this is just a very quick overview of Qpilot.
Jessica
if you want a detailed look into Qpilot, you can look into our webinars exclusively on Qpilot.
Jessica
You can reach out to us for a more detailed and live demo of Q […]
Jessica
Next and lastly, I just want to quickly touch upon our app performance module, which is another key, you know, exciting feature here at the where you'll be able to track the app level metrics, right?
Jessica
So, you will simply connect to a device and in Right, and click on this record option here.
Jessica
So, once you click on record, the app that you have selected will get installed and you simply go go about doing your typical user journey, right?
Jessica
So, this I'm showing it manually, but this same thing can be triggered through automation as well.
Jessica
At this point, you're not worrying about all the data collection that's happening in the behind the scenes, right?
Jessica
But, behind the scenes, the data collection engine is picking up on all those key metrics.
Jessica
But, we're simply going ahead and doing mimicking a user journey.
Jessica
In this case, I'm adding an item to the cart.
Jessica
And once it's done, you can go ahead and release the device and get into the reports.
Jessica
So, once you release the device, the report will be available in the report section.
Jessica
This is the summary section.
Jessica
Firstly, on the left-hand side, you'll be able to see all the relevant AML insights.
Jessica
So, the in terms of the janky frames that was the memory thresholds, any large file transfers that could be hindering your performance.
Jessica
All right, along on the right-hand side, you'll be able to see the app related insights, like your you know, your peak CPU usage, right?
Jessica
Your app launch time, how much time it took for your app to launch and become interactive, right?
Jessica
And you can deep dive into each of these sections.
Jessica
So, in the app insights, you'll be able to see we have a 20 plus time series metrics, right?
Jessica
So, you can see here that we have system level, app level, CPU utilization and you can hover over any point to see the metrics of the same.
Jessica
And on the right-hand side you'll be able to see what was happening at the UI level corresponding in your application as well.
Jessica
So, the different types of memory, latency, network, you know, transmitted and received along with battery, voltage and temperature.
Jessica
And you can also hover over at any point and you'll be able to see all the metrics correlated in a single pane of view as you can see here.
Jessica
Imagine trying to, you know, correlate and get all these different metrics from different sources, right?
Jessica
So, with Pcloudy app performance you will simply need to mimic the user journey and you'll get all these metrics in a single pane of […]
Jessica
Next, we will move into the UI insights.
Jessica
So, in the UI insights you'll be able to see, you know, all the the frame timelines, the janky frames and so on.
Jessica
You'll also be able to see the page transition.
Jessica
So, from every page transition, right?
Jessica
Like, for example, the home page to the login, sorry, the login page to the home page, right?
Jessica
The home page to add to cart, any such options like that, right?
Jessica
You'll be able to see how much time, what was the duration it took for each of the page transitions.
Jessica
And if it was, if it exceeds more than 3 seconds, it will be marked as a slow page load, right?
Jessica
So, you'll be able to see and corresponding on the right-hand side you'll be able to see what was happening in the application and which page transition took a longer time.
Jessica
So, you'll be able to, you know, give that feedback to your developers.
Jessica
And I also want to highlight that all our Pcloudy reports are easily shareable.
Jessica
They do not need to be Pcloudy users.
Jessica
Another thing that George was pointing out in terms of, you know, things taking time with, you know, feedbacks being delayed.
Jessica
These reports would really help in ensuring, you know, that there's faster collaboration and transparency between the different teams, between your development teams and so on.
Jessica
And once, yeah, the next is our network page.
Jessica
So, here you'll be able to look into every single API call that was done, the status of the API call.
Jessica
You can also filter out what were the slowest endpoints, the large transfers, and if there was any failed endpoints, that'll also be shown here with several metrics.
Jessica
And it will you can click individually into every API and you'll get all the header related information as well.
Jessica
So, for example, you're clicking into a particular API, you'll get all the header information here.
Jessica
So, this also can be useful in order to share with your dev teams and relevant stakeholders.
Jessica
And finally, we get into the log analysis, right?
Jessica
So, here you'll be able to see it will go through the entire device logs.
Jessica
So, the device logs will be downloadable here, but log analysis engine will also go through the logs and classify it into different issues.
Jessica
So, if there was any crash, all the crash log lines will be visible here and you simply need to click on it and that those relevant log lines will be visible.
Jessica
For example, as you can see here below, right?
Jessica
Similarly, for the crash, it will categorize it into these relevant issues.
Jessica
So, this is the you know, I'm just giving a overview.
Jessica
Again, reach out to us if you want a more deep dive detailed demonstration of this, right?
Jessica
But, we have 60 plus metrics.
Jessica
We're capturing both for Android as well as iOS, several metrics, and this is a key breakthrough that we've been able to achieve and that's been a big differentiator here at Pcloudy.
Jessica
So, uh really, uh you know, uh if you're interested, please do reach out to me and uh to Pcloudy and we'd be happy to help you get a deep dive demo of these features.
Jessica
So, thank you so much and uh with that, I'll be handing it over back for the Q&A […]
Jeroline
Yeah, uh that was a fantastic session.
Jeroline
Uh thank you so much, George and Jessica.
Jeroline
So, let's jump into the Q&A segment.
Jeroline
So, if you haven't already asked your questions, you can do it now.
Jeroline
Uh please drop your questions into the Q&A box at the bottom of your screen and we'll pick it up.
Jeroline
Uh here's a question for you, George.
Jeroline
So, how good is the quality of automation code that is generated by coding assistance?
George
Okay.
George
So, basically uh if you're trying to use maybe uh GitHub Copilot uh for example, and you're actually writing a prompt to generate an automation script, I would say like uh for example, let's take an example of maybe generation of script maybe using Playwright or maybe So, the the the the script that is generated is definitely comparable.
George
It may not have any kind of runtime errors.
George
And there's like a 50% chance that it may not work primarily due to locators.
George
In the sense like uh for example, if it's the link if you're trying to click links on the first page or if you're trying to interact with uh controls on the first page that you've actually given as input, it works well.
George
But, if you want to actually if your test case or the scenario that you've given actually has to navigate to a different page which is not visible uh from the home page or the page that you've given, it may not be able to learn the what I would say object properties and it may actually assume uh a lot of things and then it generate test case.
George
So, there's like 50% probability that the code might work or the script may actually work or pass or fail uh in the right way uh for but if you're actually using coding assistant, I would say like the script even if you even if the script fails, you may have to spend maybe uh 15 to 20 minutes or max of 30 minutes to fix it if it's like a simple to medium complex kind of script.
George
So, uh uh uh so, what I would like to convey is basically like it saves a lot of time.
George
You're able to generate automation scripts or at least I would say 50% of your automation script.
George
It saves a lot of time because I would say it test script development is an effort that is very effort intensive.
George
It is the effort that you spend on writing automation scripts is almost three times than writing a manual test case.
George
So, it saves a lot of time.
George
Second thing is even if you're not a highly skilled programmer or an automation engineer, you can still develop automation scripts with the help of coding assistant.
George
So, I would say the quality is not bad.
George
I would say they are good, but you may not actually get a completely working script, but things might improve because all of these AI solutions are actually improving.
George
So, I'm hopeful that maybe by end of this year or early next year, we may reach to a state where the generated automation scripts can be directly executed without any kind of errors.
Jeroline
Uh any more questions?
Jeroline
Uh there's one more.
Jeroline
What are the challenges that one would face while implementing AI-based testing solutions?
George
Okay.
George
So, one is basically like I would say the cultural resistance is definitely a problem, whether it is a testing solution or a development solution or even a business uh an AI solution that is solving a business problem.
George
So, the definitely there will be a lot of resistance from within the testing team to adopt an AI-based solution because there is always a worry about losing jobs or positions within the company.
George
So, definitely there will be a lot of resistance that you might face when you're trying to implement an AI solution.
George
You can definitely uh you should definitely involve your testing team in any kind of evaluation that you're doing.
George
And also give them some challenging work.
George
Like if I'm doing tasks like maybe like test execution can be automated or test design can be automated.
George
There are a lot of other things that they can do which requires what I would say human thinking.
George
So, you should actually give an assurance to your testers, definitely.
George
Second thing is mostly the AI solutions will definitely require I would say data uh for training.
George
If you're looking at uh maybe an execution-based uh solution which takes looks at your historical data, you may need to train the engine.
George
Or if you're looking at the user story engine which is very specific to your domain or your application, you may have to actually do a lot of training.
George
So, you may need to have high volumes of data as well as the quality of data needs to be good.
George
So, data uh I would say is very important and many organizations do not provide data and the quality of output that you get from these AI models will also be bad because the data was not The quality of data was not good.
George
Third thing is basically like don't expect results overnight.
George
AI solutions will take for some time to learn and improve.
George
So, definitely you need to give it time.
George
So, do not commit to maybe 1 or 2 weeks of implementation and then your management or leadership will start asking for results.
George
So, you may need to give it some time to learn and improve.
George
So, definitely it is not uh I would say an activity that you can that can be very much time-bound.
George
It may slip because the AI solution may require additional training and data to improve.
George
And even if you're looking at the fourth challenge, basically like if you're looking at a commercial solution, you need to have a good governance model around uh the tool implementation.
George
You may need to constantly look at the features that are coming inside your tool.
George
You need to evaluate it, provide feedback to your vendor and then improve it.
George
So, there's a proper review and the governance mechanism that should be there.
George
And many organizations do not have such a model.
George
And once you implement the tool, the tool may actually improve uh over a period of time.
George
There might be new capability, but the testing team may not be aware of any of those.
George
So, governance model is also a governance also a top challenge from based on my experience.
Jeroline
Uh there's uh one more question.
Jeroline
So, how can AI be used to conduct test impact […]
George
Okay.
George
So, there are two ways in which uh you can actually do test impact analysis or maybe two or three ways in which you can do test impact analysis.
George
If you actually have uh access to your code base, you can definitely look at what has changed within your code, and then look at what kind of test cases that you need to Or if you don't have access to the code, there are solutions that can actually look at methods uh from your binaries, and then look at what are the methods that development uh team has actually modified, updated, or deleted.
George
And then you may actually be able to find out what kind of test cases you may need to write newly for newly added methods, or you may also need may you may be able to find out the test cases that you may need to execute because your developer has actually modified, or uh what I would say deleted uh some piece of uh or some methods from your program.
George
So, that is basically the first two methods are basically one of them based on code, the other one is basically based on binary.
George
If there's I would say it's mostly I would say like looking at the code and or the or the methods, and then deciding upon what needs to be executed.
George
There's also one more method in which you can do test impact analysis, uh which is primarily uh using I would say a solution like uh maybe uh looking at the historical test runs.
George
You can actually look at the type of test cases that are executed in the past, what kind of defects were uncovered, how long did it take to run.
George
So, a lot of metrics data can be collected from your previous historical uh previous data or historical runs.
George
And this data can be used in order to decide upon what type of test cases or what test cases you may need to execute for a new build.
George
So, this is a completely AI-driven solution with a lot of training that may be required to actually train your engine based on your historical runs.
George
And these kind of solutions will actually help you to save a lot of time with respect to executing your tests.
George
So, you may be executing a low volume of test cases, but these low volume of test cases might be good enough to find majority of your defects.
George
So, these are the three types of solutions that I've come across uh for test impact analysis.
Jeroline
think there are more questions and we're already running out of time.
Jeroline
So that brings us to the end of today's session and a big thank you to our speaker George for sharing such valuable insights and to all of you who joined us today for being an engaged and interactive audience.
Jeroline
And if you have more questions and if we couldn't answer it, please don't worry.
Jeroline
You can always reach out to us at or connect with us over LinkedIn.
Jeroline
And also as a quick reminder, this session is being recorded so we'll share the link with you soon after the webinar and feel free to revisit the discussion or share it with your colleagues as well.
Jeroline
So stay tuned for our upcoming webinars and be sure to follow us on LinkedIn for more updates on AI driven testing automation and industry industry best practices.
Jeroline
So until next time, happy testing and have a great day.
Jeroline
Thank you.
Jeroline
Thank you.
Jeroline
Thank you.
