1:1 AI Mentorship · depth: one person

Learn to run the harness yourself.

Weekly 1:1 sessions on your own real work: the decision in front of you this week, worked with the method we run our own company on. Not a course. Not coaching.

One loop, four weeks, on your problem.

Models you'll learn to harness
  • Claude
  • GPT
  • Gemini
  • Llama
  • Mistral
  • Qwen
  • DeepSeek
  • Whichever ships next
Tools we get you fluent in
  • Claude Code
  • Codex
  • GitHub Copilot
  • Cursor
  • Windsurf
  • GitHub
  • Notion
  • Playwright
The problem

Your people are already using AI daily. Are you?

AI is becoming the baseline professional skill spreadsheets were twenty years ago. Someone on your team is already fluent in it. The question is whether you learn it from someone who builds these systems for a living, or from whatever a vendor's demo shows you.

0

of organizations see no significant earnings movement from AI, despite record spend

McKinsey, The State of AI in 2026

0

difference in tokens per solved task from harness choice alone, same model, same tasks

Vats & Golev, Sentient Labs, arXiv:2607.22585, 2026

The solution

Learn it on your own work, not a sandbox.

We work inside your actual job: the tasks you do every week, the tools you're evaluating, the parts of your role you could hand to an agent. You get hands-on in Claude Code, or Codex, or Copilot, whichever fits, and by the end you can tell what's genuinely useful from what's just well marketed.

Included

4-week sprint · one session a week

  • Weekly 1:1 session, 60–90 minutes, recorded
  • Hands-on work in Claude Code (or Codex, Copilot, whatever fits your setup) on your real tasks
  • An honest read on any tool or claim you're evaluating: what it actually does, not what the demo shows
  • Your own workflow redesigned around agents, with rough time estimates
  • License to use everything built during the sprint

Not included

  • AI tool licenses and usage costs (billed to you directly)
  • Building production systems for your company (that's AI Systems)
  • Support outside the agreed session cadence
Examples

What a session looks like.

Three session tracks we run most, on your work, not a sample one.

Track 01

Get fluent in Claude Code

We sit down with whatever you're actually working on this week and work it in Claude Code directly. When you get stuck, that's the lesson: we unblock it together and you see exactly why it broke.

Typical track · not a client record

Track 02

Redesign your role for agents

We map what you actually do in a week, find the parts that are genuinely repeatable, and build the harness for them. You keep the judgment calls; the agent takes the parts that were never a good use of your time.

Typical track · not a client record

Track 03

Real or hype?

You bring a tool, a vendor pitch, or a claim you saw online. We test it on your actual work instead of a demo, and you get a straight answer: worth adopting, worth watching, or noise.

Typical track · not a client record

stepmodel callhuman in the loopcheck / gate
How it works

Four weeks, one loop.

01

Before week 1

Intake

You send what you're actually working on: the tasks, the tools you're curious about, the parts of your job that feel repetitive. We pick what matters most this month.

02

Weeks 1–4

Weekly session

60–90 minutes, recorded, on your live work. We apply the method in front of you and you take it from there.

03

Between sessions

You practice, we review

You work in Claude Code (or whichever tool fits) on your own between sessions. We review asynchronously and flag what's off before it becomes a bad habit.

04

Week 4

Handoff

You leave with a redesigned workflow, real fluency in the tools, and a method you run without us. Continue monthly if it's useful. No obligation.

Talk to us

Bring the work you actually do.

Tell us your role and what you're trying to get AI to actually do for you. We reply within two business days with whether a sprint fits, and what it would cover.

  • No curriculum: your real work is the material.
  • No public pricing: the sprint is scoped to you.
  • No dependency: the point is that you can do this without us.
We don't share your details. No newsletter.
FAQ

What people ask before they write to us.

Is this a course, or coaching?

Neither. A course teaches generic material; coaching works on you as a person. This works on your actual job: this week's real tasks, with a method you keep. Sessions are recorded because the point is the work product, not the conversation.

Do I need to be technical?

No. You need to own real decisions about how your company uses AI. The method is about what the system needs (context, tools, access, checks) and who decides, not about writing code. If you do write code, the sessions go deeper.

Can a colleague join?

Yes, one additional person at no change to the sprint. If you want the method installed across a whole team or area, that's Team AI Consulting.

What happens after the four weeks?

You keep the recordings, the redesigned workflow, and the method. Some people continue with a monthly session; most don't need to. If something you built needs to become a real production system, that's a separate, scoped engagement, never assumed.