A 4-hour, in-person AI Agent Build Intensive in Raleigh. You’ll build a real research agent, learn the patterns to build any agent next, and leave owning the work. No platform lock-in. No consultant on retainer.
A working research agent you built yourself, plus the patterns and judgment to build any agent next. Concretely, by 3:00 PM Saturday you will have:
SKILL.mdWritten for a workflow you actually do every week, not a demo scenario.CLAUDE.mdSo Claude knows who you are and what you do every time you open it.You leave owning your build. No platform lock-in.
Three audiences keep showing up to these sessions. If you see yourself in one of them, the seat is for you.
You run ops at a 5–30 person service firm, agency, contractor, or manufacturer, and you keep watching AI demos and wondering where to start in your actual business.
You work in healthcare, legal, or financial services and want to learn agent-building from someone who treats HIPAA, SOC 2, and audit posture as design constraints from the start.
You’re a consultant, analyst, or internal AI champion who can prompt fine and wants to ship a real agent next, not another demo, not another deck.
Common to all three: You’ve used Claude before. You’re comfortable enough with a laptop to follow along. You can describe a process clearly. Coding experience is not required.
A 20-minute workflow audit before a single line of prompt is written. Then we build, observe, and revise, all the way through to a working agent on your laptop.
SKILL.md anatomy, the description field as a trigger, semantic matching.Four working principles, lifted from the way Cohort 001 actually unfolded. These are the mantras you’ll hear at the whiteboard, then again at the laptop, then again in the bug-hunt.
We build for your workflow, not ours.
Most AI courses teach you to copy a generic example. This one teaches you to build for the work you actually do. Your sources, your tools, your compliance posture.
We de-risk through iteration.
You leave with an agent you broke and rebuilt in the room. If it doesn’t work by 3 PM, we keep going until it does.
We transfer ownership, not dependency.
No platform lock-in. No consultant on retainer. You keep the build, the templates, the patterns, and the judgment to build the next one.
Stop automating the wrong things.
Most people automate noise. The intensive opens with a 20-minute workflow audit so the first agent you build is one worth building.
Jon Klinger · Founder, KITSCoJon Klinger teaches operators and clinicians how to build Claude agents that survive contact with real work. Cohort 001 was four students from four very different fields: a bathroom contractor, a leadership coach, a software engineer, and a hospitalist.
His method has shipped production agents in regulated clinical settings, small-business operations, and professional services. He’ll be at the whiteboard for the first hour, and beside your laptop for the next three.
Pull quote from Cohort 001. Full case study publishes May 29. (Photo pending Chi’s written sign-off.)
Thanks again for yesterday’s teaching session on AI agents. It was excellent, and the templates and slides you sent through have already given me a working roadmap. I’m putting your instruction into practice and starting to build four agents of my own: Research, LinkedIn Post, Travel Arrangement, Executive Assistant.
Anything we hedge on, we say in plain English: “Anything under 3:1 ROI we don’t build.”
A pre-class setup checklist goes out 5 days before. If you hit any environment snag, you’ll have time to fix it before Saturday.
Coffee, water, and snacks provided. Lunch is on you, there’s great food walkable around 3800 Glenwood.
If yours isn’t here, email us. We reply same business day, no autoresponders.
Yes, with care. The intensive teaches HIPAA-aware patterns and the data-separation principles that apply to any regulated workflow (PHI, PII, financial data, privileged communications). Cohort 001 included a hospitalist who is now shipping agents in a clinical setting.
This page is not legal advice. Your specific deployment should be reviewed by your compliance officer or counsel. We give you the patterns and the judgment to have that conversation well.
No. You need to be comfortable on a computer and able to describe a process you do regularly. The course teaches you to write skill files (plain English instructions in a structured format), not Python or JavaScript.
Cohort 002 is taught on the Windows + Claude Cowork + OpenClaw stack. We’ll do a Mac-native cohort in the future. Join the Cohort 003 waitlist below if that’s you.
A research agent is the highest-leverage starter agent. It teaches you skills (procedure), memory (so it remembers what you care about), and tool use (browsing, file access) in one build. Master it and you can build any agent next.
That’s the design. We allocate a full 20-minute bug-hunt block, and Jon stays after the session for anyone whose environment isn’t fully working by 3 PM.
Full refund if you cancel by June 13 (14 days before). 50% refund between June 14 and June 20. After that, no refund, but your seat is transferable to someone else in the Triangle.
3800 Glenwood Ave has on-site parking. Detailed venue notes (entrance, suite 150 directions, parking) go out with the pre-class checklist.
Yes. Date and venue announced after June 27. Join the waitlist below to hear first.
Payment is processed securely. You’ll receive a confirmation, the pre-class checklist (5 days before), and venue details by email within 5 minutes of purchase.
Drop your email and we’ll let you know the moment Cohort 003 opens. You’ll also get the Chi case study and a short essay on the “stop automating the wrong things” framework.