Matt Berman

AI Reviewer — Independent

Profile

Matthew Berman runs one of the most-watched AI review channels on YouTube — Forward Future, which as of 2026 has crossed half a million subscribers and tens of millions of views. When a new model drops — a Claude release, a Llama update, a random Chinese open-weights dump — his reaction-and-benchmark video is usually live within hours. That speed is the product: if you want to see the new thing actually running before you burn an afternoon on it, Berman gets there first.

Before the channel, he was a founder. He co-founded Teamly out of 500 Startups in 2012, then built Sonar Technologies in 2014, a messaging-automation SaaS that grew to ~25 people and was acquired in 2019. He kicked off the current AI-focused work in early 2023, which now spans the YouTube channel, the Forward Future newsletter, angel investing, and consulting. The operator background shows: his videos are less “here’s a cool demo” and more “here’s what I could actually get it to do, in roughly the time you’d give it.”

His signature format is the rubric test. He runs each new LLM through a repeatable gauntlet — coding problems, logic puzzles, the famous “how many words in your next response” self-reference test, an ethically fraught request, a censorship check — and grades them side-by-side. It is not a rigorous academic benchmark and he does not pretend otherwise. It is a consumer-grade sniff test, and that is exactly its value: you can watch ten minutes and know whether a model is worth downloading.

For developers learning AI, Berman is best used as a filter, not a teacher. The channel is where you go to decide what to try next — which local model, which new coding agent, which autonomous framework of the week. Once you know the thing is worth your time, go read the paper, the repo, or someone like Simon Willison for the deeper take. Used that way, the rapid-fire review cycle becomes a genuine productivity tool for anyone trying to keep pace with the field.