What I do
Three things, every day. All three are flavors of the same underlying question: how do you make AI actually useful?
AI that does a job end-to-end: not chatbots that answer questions, but agents that take a task, break it into steps, call tools, handle errors, and deliver a result. Voice agents on live calls. Research agents that synthesize reports. Workflow agents that replace a whole job function.
When does an agent work? When does it fail? How do you know before you ship?
Foundation models (Claude, GPT, Gemini) are the engines. The interesting work is the plumbing: retrieval architectures, prompt caching, agent orchestration, evaluation pipelines, tool-use error handling.
My takeaways live in Prompting for agents and the Patterns section.
Founders and operators ask me to look at their AI architecture. The goal is always the same: an honest second opinion. Am I overbuilding? Am I overselling the AI? Is there a simpler path?
If you want that, get in touch.
The through line: all three are about the product layer. Not training models. Not pushing benchmarks. Just making AI do real things for real people.
I build one system at a time, flat price, in accounts you own. Start with the workflow that eats the most hours.
Start with one workflow