A repeated decision or handoff
Reading and sorting documents, source-linked research, drafts that need a person to approve them, internal tools that cut copying and re-keying, and small agent workflows with known tools and clear stop rules.
Consulting
Have a problem you think AI could help with? Tell me about it.
Describe a task, decision, or handoff that takes more time than it should. I read every submission myself and reply with an honest take: whether AI fits, what I'd try first, and when the better answer is to build nothing.
01
You describe it
A few sentences about the task and where it slows down. Rough is fine; I don't need a spec.
02
I read it
Every submission comes to me. It doesn't sit in a queue someone else screens.
03
I reply
You'll get an honest read: whether AI fits, what I'd try first, or why I wouldn't build it.
04
We decide together
If it's worth a next step, we'll talk about a small, scoped proof. If it's not, you've spent a few sentences and know where you stand.
Describe the problem.
A few plain sentences are enough. Tell me what happens today and where it slows down. Please leave out anything confidential.
01What to bring
Small, repeated problems work best.
The best problems to send are small, repeated, and owned by someone who can tell good output from bad.
Sometimes the honest answer is no
If a decision has no review path, the data can't be used safely, or an existing product already solves it, I'll say so. I won't call a demo production-ready because it worked on three easy examples.
Start with the work
Describe what happens
today.
Tell me who does the work, how often it happens, and where it slows down. I'll reply with the first question I'd test.
Describe the problem ↗