The people who do the work
Testclub was founded in 2016 to test software in places where getting it wrong is expensive: trading platforms, insurance, energy and government.
More than twenty years of that work taught us where defects really come from. Rarely bad code. Almost always the things nobody specified. And the same gap that lets a bug through will let an attacker through.
That is why we now test and evaluate AI systems, the agents built on them, and the APIs behind them. The work is the same. The consequences are not.
The team
FOUNDING DIRECTOR
Omar El Dali
Omar's early career started in engineering, where he spent the next decade noticing that the most expensive failures were never really in the code. They lived in the gap between what was asked for and what was built. He founded Testclub in 2016 to work in that gap full time, testing systems for trading platforms, fintech, insurance, health, energy and government.
When generative AI arrived, the gap moved and so did he. He was testing retrieval-augmented generation (RAG) systems in the early days of generative AI, and now leads Testclub's AI evaluation and security work: finding how AI systems and agents fail, measuring how often, and testing whether they can be made to misbehave through the APIs they act on. Alongside client work he researches new evaluation methods, so they're proven before they reach an engagement.
COO
Sara Leatherland
Consultancies are judged on their findings and kept on their delivery. Sara is why Testclub gets kept. She runs how the company works: scoping engagements so they start clearly, keeping delivery tight while they run, and making sure what was agreed is what arrives.
She has spent the last three years in senior roles inside AI businesses, as COO of an AI solutions provider and fractional COO at an AI transformation consultancy. With clients, she leads on AI strategy and governance: where AI should be used, who is accountable for its outputs, and what evidence a leadership team or board needs before they stand behind it.
Values
Evidence over assertion. Every claim reproducible, every finding shown with impact.
Specification thinking. Asking questions the team may not have asked.
Discover, specify, interrogate, report. Conclusions follow from evidence, never precede it.
This way of working has a longer argument behind it. Why specification gaps are where bugs, breaches and AI failures live, and why the skill of finding them transfers.
Track record
Before security, Testclub spent years building and repairing QA practices for clients including Schroders, EDF Energy, esure, Nephila Capital, Nando's and Sportingtech, including a full QA transformation at Sportingtech. That heritage is why the security work reads unfamiliar systems quickly. We have spent more than twenty years learning how teams actually specify, build and break things.