Your engineers have AI tools.
Now give them AI skills.
Hands-on training for teams learning to use AI well, extend it safely, and operate agents in production. Taught by an engineer who builds these systems, not a slide deck.
A practical path from using AI to extending it.
Six core modules across two tracks: use AI effectively, then connect it to real systems and production workflows.
The Harness: Web, Cloud, IDE, CLI
Choose the right AI surface based on access, context, review needs, and where the work runs.
Context: Prompts, Skills & Memory
Turn prompts, project rules, examples, and skills into repeatable team knowledge.
Security Boundary: Isolation & Permissions
Understand what the model can see, what the harness can do, and where approvals belong.
Write Your Own MCP Server
Expose internal systems through safe, model-friendly tools and resources.
Agents, Tools & Workflows
Design agentic workflows that combine tools, code invocation, state, review gates, and handoffs.
Agent Monitoring & Tracing
Trace prompts, context, tool calls, costs, evals, and failures in production workflows.
Fine-Tuning
Decide when model customization is worth the data, cost, and operational complexity.
MLOps for AI Products
Manage datasets, evals, releases, monitoring, rollback, and governance.
Delivered where your team works
Workshop intensives
One to three days at your office. Whole-team immersion with labs run against your own codebase and use cases.
- Up to 20 engineers per cohort
- Curriculum tailored to your tools, stack, and security model in a pre-workshop discovery call
- Best for kicking off an AI initiative or leveling a whole org
Distributed cohorts
Half-day sessions over two to six weeks, scheduled around sprint work. Exercises land between sessions so skills stick.
- Same labs, same materials, distributed pacing
- Recordings and reference repos for every session
- Best for distributed teams and steady skill-building
Kurt Harriger
Kurt's purpose is to create clarity for others: understand what success looks like, make clear recommendations, and build feedback loops that show whether the work is on track.
He brings that lens from 25+ years in production software: EC2 deployment services at AWS supporting 5M+ hosts, Confluence Analytics at Atlassian processing 60M+ events a day, and data platforms at Optum and Rally. Today he builds agentic systems end-to-end: custom MCP servers, Claude and ChatGPT connectors, and human-in-the-loop pipelines that automate real work.
Engagements
Scoped to your team after a discovery call. No public pricing — every engagement is tailored.
Foundations
The Using AI track for teams standardizing how engineers work with web, IDE, and CLI AI tools.
- Modules 01-03
- Harness selection guide
- Reusable context and skill starter kit
- Security boundary checklist
Agent Engineering Intensive
The full core curriculum: use AI well, extend it with MCP servers and agent APIs, then instrument the result.
- Modules 01-06
- A working agent pipeline, built by your team
- 30-day follow-up session
Production Partner
The intensive plus advisory while your team productionalizes its first agents: architecture reviews, monitoring, evals, and incident support.
- Everything in the intensive
- Monthly architecture reviews
- Async access between sessions
Where is your team stuck with AI?
Share your team size, stack, current AI tools, and what you're trying to automate. I'll follow up to understand the fit and shape a training plan around your goals.