Hands-on AI for executives and teams
I work hands-on with executives and teams to turn the workflows already in their heads into AI systems they can build, inspect, and improve themselves.
We start with a real piece of work: reviewing companies, tracking a market, preparing for clients, qualifying opportunities, producing a recurring brief, or making a decision that follows a repeatable process.
You bring the judgment. We build the system around it. No coding background required.
Experienced people already operate systems. They know what to look for, which signals matter, what needs another source, and when something feels wrong.
Much of that system still lives in browser tabs, documents, spreadsheets, and memory. We make the useful parts explicit, then give the process tools, memory, repeatability, and an interface you can use.
The goal is not to remove you from the loop. It is to make your judgment easier to apply.
Usually three focused half-days, scheduled close together. We map one real workflow, build the first version live, test where it fails, and decide what should remain human.
You leave knowing how it works, what to trust, what to verify, and how to keep changing it.
An extended working session followed by clinics around individual projects and obstacles. Build one shared process or several workflows drawn from each role.
The outcome is a working prototype and a repeatable way to build—not a tour of AI tools.
If the work is valuable and repeated, it is a candidate. It does not need to sound like an AI project.
We build live. You see the prompts, tools, sources, costs, and failure modes as they take shape.
Outputs that matter get evidence, tests, limits, and review points. Consequential decisions stay human. Existing security and data policies are design constraints from the beginning; public or synthetic material can stand in when confidential data is out of bounds.
A useful system should be understandable, controllable, and honest about what it cannot do.
I build and operate AI products end to end. Current work includes Sonitura, a local screen-aware voice and computer-use assistant; burnban, an open-source meter and circuit breaker for AI spending; nextbig.dev, an intelligence pipeline reading more than 300 sources; and vote.direct, a live voting platform with verification and audit controls.
I bring the same method into an engagement: understand the process, build the useful version, test it, and leave behind a system that can keep improving.
Recent work includes general partners and managing directors at leading investment firms, and executives building companies from seed through Series A. Client work is confidential.
Tell me about one thing you repeatedly research, compare, write, track, or decide. That is enough to begin.
oday@op8.ai → Remote and in-person engagements by arrangement.