Tom Charman, Blok: Why AI Agents Will Replace Traditional Product Testing

<p>Humans tell you what you want to hear. So do the LLMs trained to be nice to you. For Tom Charman, that is the trap most synthetic research walks into, and the problem he built Blok to solve.</p><p>Tom is the co-founder of Blok, a simulator for product teams. As he describes it, building is getting faster and cheaper, so the hard question is no longer how to build. It is whether you should build the thing at all. Blok puts agents that represent real customer segments directly inside your product flow, then shows step by step how those users behave, where they feel friction, and where they drop off.</p><p>Tom and Gabriela Naumnik get into the thesis underneath the product: that experimentation is shifting from reactive to proactive, and that most companies will never have the data to do it the old way.</p><p>What Tom covers in this episode:</p><p>- Bottom-up vs top-down, and why agents interacting with the product beat a population answering questions<br>- The 87% fidelity benchmark Blok reports and what it takes to trust simulated results<br>- Why regulated industries in finance, healthcare, and government came first for Blok<br>- How discovery can drop from four to six months to one or two days<br>- Why adaptive interfaces could make simulation the only path to statistical significance for 99.9% of companies</p><p>For founders, product leaders, and operators deciding what to build next.</p><p>highstakes.fm</p>