Field Notes
Field notes from real builds, not vendor marketing.
I have published roughly forty long-form articles, taught this material to executive cohorts, and shipped software as someone who is not a developer. Those essays are chronological and first person. They answer what happened to me.
This is the other axis. Topical, durable, and built to be extracted. Every chapter answers a question, opens with the answer, and shows the published work it came from in the margin. If a model reads one paragraph of it, that paragraph should still be correct on its own.
9 chapters
- 01
Context Engineering
How do you give an AI agent the context it actually needs?
Context engineering is the continuous work of deciding what an agent sees, and when. It runs through every phase of a build rather than just the first one, and it is where most agent work quietly fails.
- What is context engineering, and how is it different from prompting?
- How do you structure the first phase of a build?
- How do you hand the context over when the build starts?
- What belongs in an agent’s context, and what does not?
- Why does an agent drift, and where does it start?
- How do you carry context across sessions and across agents?
- Does any of this change if you cannot read the code?
- 02
Agent Memory and Second Brains
How do you give an AI agent memory that survives the session?
An agent forgets everything when the window closes. A second brain is the durable half of the system: structured, linked notes the agent reads back on demand, so accumulated thinking outlives any single conversation.
- Why does an agent forget, and why is that hard to fix?
- What kinds of memory are there, and which one do you actually need?
- What is a second brain, in concrete terms?
- How does the model actually reach the notes?
- What changes once your thinking outlives the session?
- What happens when more than one person contributes?
- What went wrong building this?
- 03
Coding Agents and Harnesses
What is the difference between a model and the harness you run it in?
The model writes the code. The harness decides what it sees, what it may touch, and when it stops. Most of the difference between a good session and a wasted one comes from the harness, not the model underneath it.
- 04
Building Without Being a Developer
Can you ship production software without being able to write the code?
Yes, with conditions, and the conditions are the interesting part. What you trade is line-level control. What you need in exchange is judgement about systems, an honest test loop, and the discipline to verify what you cannot read.
- 05
Orchestration and Multi-Agent Work
When is more than one agent worth the trouble?
Rarely, and then decisively. Most tasks are faster with one agent and a clear brief. Fan-out earns its keep when the work is genuinely independent, or when you need several honest opinions before you commit.
- 06
AI Sovereignty and Open Weights
What does it mean for a European organisation to control its own AI?
Sovereignty is not a slogan about hosting location. It is a set of concrete answers about where the weights run, who can revoke your access, what leaves your network, and what you would do if a vendor changed terms tomorrow.
- 07
Organisational Design for the AI Era
How should a company reorganise itself around AI, if at all?
Most AI initiatives fail on org design rather than on technology. The work lands in a team with no mandate, no budget for failure, and no route to the people whose processes would have to change for any of it to matter.
- 08
Verification
How do you know an agent's work was right if you did not watch it happen?
Agent failures are quiet. The work completes, reports success, and leaves something broken underneath. Verification is the discipline of making that gap loud, and it is where most of my own worst findings came from.
- 09
What I Shipped and What It Cost
What have you actually built, and what did it take?
The products I built and run, described in terms of what they do and what building them cost me. No architecture diagrams and no capability claims I cannot point at a published source for. Just the account.
Take the corpus
The whole thing is a download, and that is deliberate.
These notes were written and kept in The Curator, my open-source second brain. It reads documents, writes an interlinked wiki of entities and concepts, and serves that wiki to a frontier model over MCP. Everything stays as markdown on your own machine.
So take the corpus. Ingest it into your own copy and ask it questions I have not thought to answer. That is a better demonstration of the tool than anything I could write about it.
- field-notes-complete.mdComplete corpus, one markdown fileEvery chapter in order, with a provenance header. Best for reading, or for pasting into a model with a long context window.
- field-notes-chapters.zipChapters as separate files, zippedOne markdown file per chapter, frontmatter intact. This is the one to use for ingestion, because each chapter becomes its own document.
Both files carry a provenance header with the source URL, the licence and the build date. They are regenerated on every deploy, so they never lag behind the pages.
Step-by-step: install and ingest⭐ Star The Curator on GitHubllms.txt