At 30% time and effort savings, AI coding is worthwhile

What new advantages apart from cost-cuts, time-cuts and automation does agentic AI bring in the realm of software testing?

Agentic AI is reshaping the future of testing. Some of the direct advantages are in quality, savings and productivity. Infusing quality at the requirement stage ensures clear, testable requirements. Innovative test ideas and mapping requirements to test cases enhance validation and traceability, driving continuous improvement for robust software. When it comes to effort saving, we can generate test cases from requirements in any format (descriptions, pictures, wireframes). We can now drive automation test cases and review bulk results with actionable insights, saving effort and enhancing efficiency.

And that ties into productivity?

Productivity booster agents achieve huge efficiency gains with ready-built agents and the ability to create custom agents using our agent builder, streamlining processes and enhancing productivity.

Developers tend to struggle with real efficiency gains, waiting-time-on-builds-and-tests, redundancy-fears and new skills when it comes to low code and AI-code. Your take?

Today teams are overwhelmed by numerous options and social feeds. While many vendors tout AI-based testing, true value realisation is something that needs to be looked at closely. In my view, basic capabilities of a real agentic testing should help us drive huge efficiency gain ensuring quality to avoid retesting.

Exploratory agents like Testing Drone help you crawl websites, run checks (latency, spelling), report results, etc.

How?

Adapting to changing priorities, create a lot of reutilisation factors to address resource availability. Simple capabilities like Shift-left testing to ensure early test case creation, wireframe utilisation, strong priority mapping, and flexibility to integrate with existing tool chains and skillsets, low/no/pro code options and purpose-built agents should deliver tremendous value for the team.

While many vendors tout AI-based testing, true value realisation is something that needs to be looked at closely.

A recent study showed that 67% of developers spend more time debugging AI-generated code, while 68% spend more time resolving security vulnerabilities. Faster code is also translating into more time needed in debugging code as well as throughput dips and delivery-stability issues. Your views?

AI offers significant speed and productivity benefits, but challenges accompany these gains. AI-generated code often lacks context and best practices, leading to not working as expected and introduces risks, often new unfamiliar codes can increase debugging time and difficulties. Balancing AI’s benefits and challenges is crucial. Despite these hurdles, in my personal view, a net 30 per cent time and effort saving makes the use of AI-generated code worthwhile.

Do agentic testing approaches affect directions like Shift-left, exploratory testing and containerisation in some way?

Agentic testing aligns with the Shift-left approach, starting with requirement quality analysis, test case creation, and automation from wireframes. This saves effort for developers, business analysts, and testers, delivering high-quality applications. Exploratory agents like Testing Drone help you crawl websites, run checks (latency, spelling), reports results etc. Agents automate repetitive tasks, freeing testers to focus on high-value work. Parallel orchestrated test execution ensures consistent testing across containers, with insights and reports shared for action.

Can agentic AI testing integrate well with manual coding and development? Or does it have to be end-to-end AI in the SDLC?

UiPath Agentic testing offers a modular approach, assisting with requirements, manual testing, test automation, test data management, analysis etc. It also performs health checks on application services in production, ensuring comprehensive and efficient testing processes.

An insurance client increased regression test coverage by multi-fold that helped reduce production defects significantly.

How strong are these alternatives on aspects like accuracy, QA, test-flakiness, code-review load, consistency, pointing/choosing tests to the right areas, false positives and tech-debt issues?

UiPath Test Cloud provides you with the accelerated speed to clear your technical debts, enhances quality to reduce production defects, and provides insights to lower Mean-time-to-Resolution. It reduces duplicates and flakiness with actionable insights. Specific agents like Stability Inspector identify flaky tests and false positives, while impact-driven testing keeps priorities on track.

Does agentic AI change the software license model in a big way—compared to Cloud/SaaS models?

Agentic Testing licensing offers flexibility for accommodating team elasticity, benefiting customers by allowing resource additions from various teams or SIs. It enhances parallel execution capabilities to speed up testing and is generally more cost-effective than traditional approaches.

Do these testing outcomes integrate well with development, documentation, DevOps and deployment areas of coding?

Our agentic testing solution is open and flexible, offers extensive integrations with various ALM, Defect, CI/CD, version control systems, webhooks, Microsoft Office, PDF, and many more.

Can they handle complexities beyond the pattern-part of tests, beyond regression tests?

Our Agentic Testing surpasses pattern-based complexities, offers higher test coverage and automation rates, saves significant effort. It excels in generating exhaustive test scenarios, handling dynamic conditions, managing locator changes, defect analysis with actionable insights, best practices compliance checking, and more.

Any examples of outcomes from the pilots/deployments you have already done?

Our team has numerous deployments across BFSI, Telco, Manufacturing, Retail, Government sectors etc., delivering significant value realisations for customers in the region. Few among the several outcomes that our customers have driven are projects like 50 per cent time saving on the entire S4/HANA migration journey for a Telco. Quarterly application releases for a bank went down to biweekly which helped them reduce technical debts significantly. Another insurance client increased regression test coverage by multi-fold that helped reduce production defects significantly as well as reduced specialised resource dependency to manage user journey testing cutting across mainframe, citrix, mobile, and web applications. A retail customer gained confidence to migrate 180 applications from one cloud to another within stated timeframe. A few BFSI customers fulfilled their regulatory compliances adopting our solutions.

pratimah@cybermedia.co.in