Course Arc
Build a multi-agent AI office, one guide at a time.
Thirteen guides, in dependency order, from a single agent-or-not decision through retrieval, guardrails, routing, and the flywheel that improves it. By the end you have a scoped, instrumented system, not a pile of demos.
The sequence
Thirteen steps, in the order they depend on each other.
- 01
AI Agents vs. Prompts: When to Use Each, and When Not To
Use one prompt for single-step work you can check yourself. Use AI agents when the next step depends on what the model finds. Here is how I decide.
August 13, 2026
- 02Scheduled
Structured Outputs and Cognitive Architectures
Getting a model to return a shape you can act on instead of prose, plus the reasoning loops built on top of it: ReAct, Reflexion, plan-and-execute.
Shipping September 3, 2026
Read Step 1 first.
- 03Scheduled
Tool Interfaces: Reaching a Database or an API
How an agent reaches out to fetch or change something: the MCP bridge, and the read-vs-write scopes that keep it from doing more than it was asked.
Shipping September 18, 2026
Read Step 1 and Step 2 first.
- 04
How RAG Answers From Your Own Documents
Learn how RAG searches your documents, where retrieval breaks, and how to choose chunks, embeddings, and a similarity floor.
August 8, 2026
- 05
How to Choose a Retrieval Cutoff
A practical way to set the similarity floor for RAG: real questions, source checks, ranked scores, and a refusal rule.
August 21, 2026
Read Step 4 first.
- 06Scheduled
Knowledge Graphs
The other way to answer a retrieval question. Sometimes a graph of entities and relationships beats a similarity search, and sometimes it does not.
Shipping September 8, 2026
Read Step 4 first.
- 07Scheduled
Routing Work to Specialists
How one request gets handed to the right agent instead of one model trying to do everything: the fan-out pattern this Office runs on.
Shipping September 4, 2026
Read Step 1 first.
- 08Scheduled
Concurrency and What Agents Can't See
What breaks when two agents run at the same time and neither one can see what the other just decided.
Shipping September 15, 2026
Read Step 7 first.
- 09
The AI Builder Stack, Translated: LangChain, LangGraph, CrewAI, n8n
Your proposal says LangChain, LangGraph, CrewAI, n8n, MCP, Streamlit. What each one is, when it is overkill, and what to ask before you sign.
August 18, 2026
Read Step 1 and Step 3 first.
- 10Scheduled
Dialog State: Remembering Mid-Conversation
How an agent holds onto what a reader already told it, and picks the thread back up after a restart instead of losing it.
Shipping September 14, 2026
Read Step 1 and Step 9 first.
- 11Scheduled
Evaluating an Agent, With No Fine-Tuning
Grading what an agent does against real questions, using the same retrieval floor and refusal rule from Step 5.
Shipping September 9, 2026
Read Step 4 and Step 5 first.
- 12Scheduled
Scaling: Real Time and More Than One Tenant
What breaks first when one agent becomes many concurrent users, built before it is written.
Shipping September 22, 2026
Read Step 8 and Step 11 first.
- 13Scheduled
Data Flywheels
Turning graded results back into better retrieval and better prompts. The loop that closes the arc.
Shipping September 29, 2026
Read Step 11 and Step 12 first.
Start a project
Ready to scope it and ship it?
Have an AI workflow, reporting gap, or system you can't seem to get off the ground? Scope it with me. You'll get a real read on what to build first and what it should cost.
I review every submission and reply with a real read on fit.