Expertise hub

Everything we know about AI and testing.

Two pillars, sixteen articles: using AI to test faster and smarter, and thoroughly validating AI systems themselves. No hype, just what we apply in practice.

// AI within the testing process

AI as a tool to test smarter

AI-assisted tooling that builds your automated testing faster and maintains it more cheaply.

AI-assisted testing

Self-healing tests: how AI repairs locators automatically

Why teams lose 40 to 60% of their automation time on maintenance, and how self-healing addresses that.

 AI-assisted testing

Three types of self-healing: selector, timing and intent-based

One label, three completely different techniques, with a large quality gap between tools.

 AI-assisted testing

Generating test cases with AI: from user story to test in minutes

Where AI test generation works well, and why generated is not the same as production-ready.

 AI-assisted testing

Synthetic test data with AI: the end of the data bottleneck

Realistic test data without GDPR risk, and what to watch for with the term "synthetic".

 AI-assisted testing

Visual regression testing with AI: more than pixel diffing

AI-based visual validation looks at what a user notices, not every deviating pixel.

 AI-assisted testing

Root cause analysis: AI that summarises failure analyses

From forty separate test reports to one usable insight into the underlying cause.

 AI-assisted testing

Exploratory testing with an AI copilot

An AI agent can explore exhaustively. Why an experienced tester still remains essential.

 AI-assisted testing

What AI testing still cannot do today

An honest overview: what is production-ready, and which signals point to hype.
// Testing AI systems

Validating AI applications as a test object

Testing strategies for LLM and ML applications: quality, robustness, bias and EU AI Act conformity.

Compliance

The EU AI Act: what does it mean for your organisation?

The deadline for high-risk AI systems shifted sixteen months. Why that isn't an exemption.

 Compliance

Conformity assessment: what test documentation you need to show

What a conformity assessment concretely requires: documentation, risk management, logging and test evidence.

 Testing

Why traditional testing doesn't work for AI systems

A fixed expected result doesn't exist with a model. Why that calls for a different testing approach.

 Generative AI

Red teaming for LLMs: what it is and why it isn't a pentest

The difference between hunting for a technical flaw and hunting for a behavioural flaw, and how a red team exercise unfolds.

 Generative AI

Hallucinations in LLMs: how do you measure it

An LLM confidently making something up isn't a bug you fix with a patch.

 Testing

Bias & fairness testing explained

A model can score perfectly on average and still systematically disadvantage a group.

 Testing

Model drift: why an AI system that works well today fails tomorrow

Validation at delivery is a snapshot. The world around it keeps changing.

 Generative AI

Testing AI agents: when a system makes its own decisions

An agent calls tools and acts autonomously across multiple steps. That calls for a different testing approach than a chatbot.
// Blog

Notes from the floor

Personal posts from our consultants: what we encounter, try out, and learn in practice.

Pricing & tools

Fly me to the moon: GPT-5.6 Luna is 80% cheaper

What OpenAI's price cut means in practice for your credits in GitHub Copilot.

See all blog posts →
// Let's talk

Looking for something specific?

We're happy to think along with what AI can concretely mean for your testing process or your AI application.

Book a call Back to aitesting.be