I build AI systems for people who have actual work to do, then write down the parts that survived contact with reality.
I run OpusJake: a library of the prompts, playbooks and skills I use on real client work, a community called Opus Lab, and a consulting practice for founders who want AI in production rather than on a slide. Most weeks split between building for clients and teaching the people who will run it after I leave.
I have been interested in AI since AlphaGo in 2016. Watching a machine play move 37, a move no human would have played, and which turned out to be right, reframed the whole thing for me. It was not brute force. It was taste.
My first real job with AI was with Schneider Electric in Japan in 2021. That was where the theory met a factory floor, a legacy system, and people whose jobs depended on it behaving. It taught me more than any paper did.
Since then I have worked with several startups and founded Beacon, an AI-native agency that helped companies show up correctly when people ask ChatGPT and Claude about them. I exited it. Now I consult and teach, helping founders and CEOs become AI fluent so they can actually use the technology instead of nodding along in meetings about it.
Agents and automations that run inside a real workflow, with the evals and the handover.
The instruction files, tools and permissions that make an agent do the job and stop.
Teaching founders and teams to build their own: Opus Lab, workshops, and advisory on what is worth building.
How a company is represented when someone asks an answer engine about it.
The content and follow-up loops a small team can actually keep running.
Agent builds, automation, or a second opinion before you commit a quarter.
Start a conversation