Talks and workshops

Testing AI Agents with Playwright and MCP: A Practitioner's Guide

45 to 60 minutes, delivered at SEETEST 2026, Ljubljana, engineers, SDETs, test leads

Agents claim to have done things. This talk shows how to find out whether they did: bounding the task, recording every tool call, asserting on end state instead of the agent's summary, and the Playwright evidence to keep for each run. Live examples with Playwright MCP, including the cases where the agent confidently did the wrong thing.

The Human Side of AI Testing

30 to 45 minutes, delivered at BCS SIGiST, London, July 2026, testers, leads, managers

"Every check passed. Ben still could not use it." What AI-assisted testing is good at, what it reliably misses, and how a team keeps a person's judgement in the loop without becoming the bottleneck. Built on stories from twenty years of testing national-scale services.

Do AI-written tests catch bugs? What the Evidence Lab found

30 to 45 minutes, new for 2027, engineering leaders and anyone signing off AI-assisted testing

Built on the published Evidence Lab investigation: the same AI wrote two suites for the same app and one caught every injected bug while the other missed eight. Why, what it means for how teams adopt AI test generation, and the one-afternoon check that tells you which kind of suite you have.

Workshops

One-hour demo (free for communities, recorded), three-hour interactive (up to 15 people), full day (up to 12)

Hands-on, using a prepared example agent and test project so every attendee runs the checks themselves. Attendees leave with the AI Agent Testing Checklist and a working sign-off check. Setup guide sent a week before; a corporate network that blocks AI tools is the one thing to check in advance. Workshop details.

Bios

Copy as is. Please do not add titles that do not appear here.

Short, 40 words

Qambar Raza (Uncle Aaroh) tests what AI systems actually do. Stanford Code in Place section leader, Microsoft MVP, LinkedIn Learning instructor with 90,000+ learners, former BBC Principal Test Engineer, 20+ years in software engineering.

Medium, 100 words

Long, 160 words

Qambar Raza, known online as Uncle Aaroh, is a Playwright and AI agent testing specialist with 20+ years in software engineering. He taught Stanford's CS106A Python course as a Code in Place section leader (2023), is a Microsoft MVP in Developer Technologies (2025 and 2026), a LinkedIn Learning instructor with 90,000+ learners across multiple Playwright and Selenium courses, and a former BBC Principal Test Engineer. He speaks on testing AI agents with Playwright and MCP, and on the evidence teams need before they trust automation. Based in Manchester, UK. His talks in 2026 include SEETEST in Ljubljana and BCS SIGiST in London, and he runs the Evidence Lab at uncleaaroh.com, where claims about AI in testing are put through reproducible experiments and published with their code and results. Through consulting and training he helps engineering leaders decide, on evidence, whether an AI system is ready to ship.

Pronunciation: Qambar, "KUM-bar". Raza, "RAH-za". Title line for slides: Qambar Raza, Microsoft MVP, LinkedIn Learning instructor, Stanford Code in Place section leader.

Recordings

How To Test This podcast, episode 40: Playwright At Scale, 56 minutes with host Mamadou N'diaye, September 2026.
Conference recordings from SEETEST 2026 and BCS SIGiST 2026 are published by the organisers; links added here as they appear. Short clips are on the Uncle Aaroh YouTube channel.

Logistics

Based	Manchester, UK. Remote worldwide; on site in the UK and Europe.
Formats	Keynote or track talk (30 to 60 minutes), 60-minute deep dive, workshops as above, panels.
Fees	Community and non-profit events: talks are usually free, travel covered. Commercial conferences and corporate events: fee quoted within two working days of the enquiry.
Needs	HDMI or USB-C, own laptop, a clicker, 16:9. For live demos, a wired connection where possible. Nothing else.
Recording	Yes, please. I ask for a copy of the recording and the right to share clips; organisers keep full rights to their own recording.
Lead time	Four weeks for a talk, six for a workshop. Shorter is possible, ask.

Talks on testing AI agents, Playwright at scale, and the human side of quality.