Discussions

With a normal API or UI you test deterministic systems — same input, same output, every time. AI agents break that contract completely. The same prompt can produce different reasoning paths, different tool call sequences, and different outputs across runs. That non-determinism is the core challenge. Ηοw would you test this system? A question that sparks the skills needed for testers in the new era! 

Ever been ghosted by a JOB? Check the resource below https://www.youtube.com/shorts/xDIf0YFC1UI

Give it a try. More to come  https://www.youtube.com/shorts/5Eay25HSuzw

A malicious file labeled “2026_Salary_Increase.pdf.exe” was opened by a large portion of employees — including the Incident Response team. Let that sink in. This isn’t about lack of intelligence. It’s about how easily human psychology can be exploited:Curiosity (salary increase)Urgency (what if I’m missing out?) Trust in familiar file naming patterns Key takeaways for teams: • File extensions still matter — .pdf.exe is a red flag• Security awareness training must be continuous, not one-off• Even security teams need realistic phishing simulations• Technology alone won’t save you — behavior will The real vulnerability isn’t the system. It’s the assumption that “this wouldn’t fool me.” When was the last time your organization tested this?#CyberSecurity #Phishing #SecurityAwareness #InfoSec #QA #RiskManagement 

A few years ago, I was proud of a UI automation suite I built.Hundreds of tests.Clean reports.Everything green.Until production started failing.Users reported errors we had never seen.But our UI tests were still passing.So I went deeper.Instead of clicking through the UI, I started testing the APIs directly.That’s when things became obvious:• The API returned inconsistent data• Some responses were missing critical fields• Error handling was completely broken• Edge cases were never validatedThe UI was masking all of it.It handled errors silently.It retried requests.It hid failures behind “friendly messages.”Our UI tests were verifying the illusion of stability.Not the system itself.That changed how I approach testing.Now I start with APIs:→ Validate contracts→ Break requests intentionally→ Test edge cases early→ Then verify UI behaviorBecause if the foundation is broken…Testing the surface won’t help.Do you start testing from the UI or the API?

It’s easy to jump too far in the future and imagine what AI will look like and how it may help us do all the things.However, in reality, we have to take steps towards a future that (may) look like that.What tedious, boring or repetitive tasks are you actually using AI for? (or plan to in the near future)

Hi everyone, I’m James, a pentester focused on web application and API security. I spend most of my time exploring vulnerabilities and real attack paths. For the past few years, I’ve been building ZeroThreat AI, an automated pentesting tool with exploit validation and agentic AI testing capabilities. I joined the community to share ideas, learn from others, and discuss how security testing is evolving from manual to automated, and then with agentic AI tools. Ciao!!!!

I once tested a password reset feature. Everything worked.Emails sent correctly.Tokens expired properly. But out of curiosity I tried something strange.I requested 100 password resets in 10 seconds. The system happily sent them all.That meant anyone could spam users with reset emails. Not a typical functional bug.But definitely a security and abuse issue.Sometimes testing means asking unusual questions like:"What happens if someone abuses this feature?" QA is not just about correctness.It's about resilience. Have you ever tested rate limits intentionally? 

One of the biggest lessons in my QA career came from a failed automation suite. We had hundreds of tests.Great coverage.Fast execution.But production still had bugs. Why?...........................Because our tests verified features, not user behavior. Real users behaved differently.They navigated unpredictably.They refreshed pages.They switched devices. Automation eventually improved when we started designing tests around user journeys instead of components. Automation should mirror how real people use the product.Do your tests reflect real user behavior?

Early in my automation career I believed something dangerous:"If automation passes, the system is safe."One release proved me wrong.All tests were green.CI pipeline passed.Deployment happened.Minutes later support reported users couldn't complete checkout.The issue?Our automation mocked the payment service.The real integration had changed.Automation verified the mock, not the system.That moment changed how I design tests.Now I always ask:Are we testing reality?Or just a simulation?Automation should reduce risk.But only if it tests real system behavior.Have you ever seen automation give a false sense of security? 

Hi Team,We are planning to conduct a PoC for a Salesforce project. We would like to understand Tosca’s capabilities for Salesforce automation, including its support for rapid automation. Can anyone share your experiences with us.Thanks,Sudha

Hey all!I would like to buffer a regex value. I have a very long payload with customer data and I’d like to extract only the customer’s number. This whole payload is stored in “Result”. I have built this regex: {REGEX["customerNumber":"[0-9]+"]}And I want to store this extraction into a buffer.Anyone can support please? Thank you!

Please see below steps taken in the screenshot.I am getting below error message. Please advise. 

Hi All,My name is Deepak Behera. I’m a Quality Engineering professional with around 19 years of experience in software testing, with a strong focus on test automation and AI‑driven QA practices. My background spans telecom and complex enterprise systems, where I’ve worked as a hands‑on QA engineer and quality leader, building robust automation suites and improving test strategies end‑to‑end. I hold a BE in Computer Science and an MBA in Information Systems, and I’m currently scaling my expertise towards QE Architect and managerial roles while also founding Vim‑Ana (www.vim-ana.org) , a startup in the UAV/drone space where disciplined testing and reliability are critical.In this community, I’m looking to both share and learn: sharing practical insights on test automation frameworks, API testing, AI‑assisted testing, and QA leadership, and learning from others on how they are evolving their quality strategies, tooling, and career paths.My LinkedIn: https://www.linkedin.com/in/hideepak/

Compare the Tricentis AI testing with other AI products in the market.Provide the advantages over other tools in the market for the automation testing using Tricentis. 

On the ShiftSync website (https://shiftsync.tricentis.com/), I keep getting logged out every other day. Even when I select “Remember Me” before login, it doesn’t work, when I open the site, the default dashboard is visible and I have to click the Login CTA again and re-enter my credentials.Is anyone else experiencing this issue, or is it just happening to me?

In many workplaces, feedback becomes a routine activity. Meetings happen. Forms get filled. Comments are recorded. But expectations stay unclear, examples are missing, and next steps are vague. Over time, feedback turns into a process to complete rather than a responsibility to grow.A more useful way to think about feedback is as a learning tool, not a policy requirement. Effective feedback is grounded in facts, considers context, makes expectations explicit, and focuses on behavior rather than personality. Most importantly, it leads to clear actions and ongoing conversations, not one-time events.This shift is especially relevant in testing and engineering teams, where intent is usually good, but outcomes often fall short. Without structure and clarity, even well-meant feedback fails to create progress.Sharing this here to encourage reflection and discussion on how feedback is given, received, and used for real growth within teams.https://ujjwalkumarsingh.substack.com/p/feedback-from-p

🚨 Security warning: 840,000+ users affected by malicious browser extensionsPCWorld reports that dozens of popular browser extensions (Chrome, Firefox, Edge) were quietly infecting users with GhostPoster malware — some since 2020.These extensions can:❌ Track browsing behavior❌ Redirect traffic & hijack affiliate links❌ Download additional malwareThey’ve been removed from extension stores, but they’re still active if you installed them before.👉 Full article + list of extensions to uninstall ASAP:https://www.pcworld.com/article/3037666/840000-users-hit-by-malicious-browser-extensions-uninstall-these-asap.html🔐 Reminder:“Installed from an official store” ≠ “safe forever.”Audit your browser extensions regularly — personally and at work.

Hi everyone, I'm currently working on integrating Generative AI into our test automation framework and could use some expert advice. Has anyone had experience using AI to enhance test case generation or maintenance? Any recommendations on best practices or tools would be greatly appreciated!

I recently completed my preparation for the ISTQB CTFL-Foundation exam, and one thing I noticed is that the topics that seem easy in theory can become confusing in actual exam questions.For example, concepts like:Test design techniques (e.g., equivalence partitioning, boundary value analysis) Test levels and types Psychology of testing Risk-based testingLook simple in the syllabus, but when presented in scenario-based questions, many candidates hesitate because the wording can be tricky.Initially, I focused mainly on reading the syllabus and memorizing definitions, but that wasn’t enough to confidently answer situational questions in the CTFL exam format. What helped most was practicing topic-wise questions and reviewing explanations of both correct and incorrect options.Another challenge many face is exam timing. If you’re not used to exam-style questions, you might spend too much time on some scenarios. My suggestion for CTFL aspirants:Try practicing questions from P2PExams in a time

Fiori application scanning of a drop down used to show the element type as ‘Dropdown’ before but now it is showing as ‘Input’ like a text field. We can scan the element, but the drop-down options are no longer showing for selection in TOSCA (which they were used to properly show before). This is blocking the scripting as entering a text in Dropdown is not possible.

Delegation of the tasks is important.This helps you to focus on important things for you in testing. Also, it helps others an opportunity to work on something different from their reglat tasks.Now, being a testers/lead/manager you might have delegated tasks to others.Which task(s) did you deleagte?And what worked well?