Not generic AI literacy. Nine ready-to-deliver sessions on the things that decide whether AI pays off: coding agents in production, harness and loop engineering, token spend, model routing, evals, and picking the right projects. Built from your tools, your workflows, and your bills.
See the catalog →Or book a free 30-minute call and I'll tell you which session fits.
Netflix made AI fluency an expectation for every role, not just engineers. Every company copying Netflix now has to define it and train it, and most have no playbook. Generic seminars don't stick because they train people on someone else's tools.
This session is the playbook, built from your stack.
Every one is built from work I do and teach in public — the CloudYeti channel is the preview. Each runs as a half-day or full-day, on your stack, and ends with a capstone on your own usage or bills.
Claude Code and agent workflows on your repos, your CI, your review process — with the guardrails that keep agents out of production by accident.
How to run agents unattended and trust the output: feedback loops, human-in-the-loop gates, work logs, and the harness files that turn a one-off agent into a system that runs while you sleep.
Caching, batching, budgets per team, and the defaults that quietly triple a bill. Your team leaves with a spend dashboard for your own usage, not a slide about someone else's.
Which model for which job, when the cheap one is the right one, and how to route between them so quality stays up while cost comes down. Commitment math included.
You can't improve what you don't measure. Build an eval set for one real workflow in the room, so "the AI got better" stops being a vibe and starts being a number.
The audit discipline as a skill your team keeps: read the AI and cloud bills line by line, match spend to output, and build the cuts-and-keeps list yourselves.
Where to invest heavily and what to kill: an ROI triage your team applies to its own backlog in the session, with kill criteria you'll actually enforce.
Most paid AI seats sit idle. Working habits, review norms, and what your team should never paste into a model — measured on your own usage report (Copilot, ChatGPT Enterprise, or Claude), so the session ends with a real adoption number.
Where the money goes, which line items are worth defending, and what to ask your teams in the next planning cycle.
Don't see the exact shape you need? The tenth session is the one we scope on a call.
"Thank you for your training. You do an excellent job explaining. I had been a technical instructor for many years — you are elite."
— YouTube comment from a fellow technical instructor. Also from students: "You explain conceptually, then show the console. Most instructors start with the console and miss the concepts." · "The most informative 15 minutes I've had in months." · "Delivery is simple, and precise."
The half-day session is $2,500, customized to your stack, materials included. Full-day and multi-session formats are scoped on the call. Remote or in person in the DC area.
This is hands-on training for a working team, built from your real workflows. It is not a generic "intro to AI" seminar, and it is not implementation work. If the training uncovers build work, that gets scoped separately.
It also pairs well with the Cost Audit: I train the team on what the audit found, so the waste does not come back.
No pressure. Tell me about your team, and I'll tell you whether this is worth your money.