Model Watch

About the show

Model Watch

Model Watch is the weekly screen-share series that does the AI testing the rest of the market is too rushed to do properly. Every episode we put a model, a tool, or a piece of software through a controlled set of tasks live on screen — the prompts, the failures, the edge cases, the cost-per-call. Nothing is staged. Nothing is the model vendor's demo loop. The host runs the same battery of tests against every contender so you can compare like with like, and we report back on which one is actually fit for the work you are trying to do that week. The result is a running, evidence-based record of which AI tools earned their place in a real workflow and which ones did not. We cover frontier model releases (Claude, GPT, Gemini, open-weights launches), the AI coding stacks, the agent platforms, the new vertical tools that promise to replace a process, and the breaking news that matters — measured against a written rubric, not a vibe. Sponsors who land here are advertising into a buying decision, not a content feed. Viewers come to Model Watch when they are about to spend, switch, or specify — they are evaluating, and they expect to leave with a recommendation. That is why this is some of the highest-intent advertising inventory APH carries, and why the companion display unit on every episode page and the editor-read pre-roll are the two formats most-bought against it.

Audience

AI practitioners, developers, CTOs, engineering managers and tech leads who specify tools for their teams — plus the early-adopter founders and product leaders who shape which platforms get rolled out across an organisation. They are senior enough to sign for new tooling, technical enough to vet a screen-share critically, and looking for an editor they can trust over a vendor demo. Roughly two-thirds are based in the UK, EU and US; the rest are split across the GCC, India and East Asia. They read with a buying budget in hand and an internal decision to make; many are running formal AI evaluations or specifying tooling for teams between ten and several hundred people. Episodes get shared into Slack, screenshots get pasted into procurement decks, and verdicts shape which platforms get the next contract.