Build real AI agent systems.
No engineering background required.

Weekly configs, failures, and fixes from a non-engineer learning agentic AI in public. For late starters who think in systems, not code.

Join the build log → Start here

Who this is for

  • Non-engineers who want to build with AI, not just prompt it
  • Operators, analysts, and domain experts with real process instincts
  • People 35–55 who feel behind but learn fast when shown the actual work
  • Anyone who wants configs, wiring, and debugging, not inspiration

Who this isn't for

  • Engineers looking for advanced tutorials
  • People who want quick hacks without understanding why
  • Anyone expecting polished, certain answers
  • Those allergic to seeing work-in-progress and honest failures

What you'll get

Practical resources from someone 2–12 months ahead, showing the work honestly.

Weekly build logs

Real agent stacks I'm building, complete with what broke and what finally worked.

Configs & templates

Actual files you can use, not abstractions. Tool configs, prompts, and wiring examples.

Failure post-mortems

Where agents break down: handoff failures, guardrail gaps, edge cases that matter.

Checklists & patterns

Repeatable approaches for debugging, testing, and iterating on agent systems.

How this works

No course. No cohort. Just honest documentation of building agent systems from scratch.

  1. I build something real Agent stacks, tool integrations, workflow automations. Using free and accessible tools.
  2. I document the process Configs, decisions, failures, iterations. What I tried, what broke, what I learned.
  3. You get the build log Weekly(ish) emails with the full story, templates attached, questions welcome.

About

I'm 48, not an engineer, and came to agentic AI late. My background is financial markets, trading ops, risk management, and large transformation projects in regulated industries.

I've spent decades working inside messy, real-world systems where precision matters, failures cascade, and "it depends" is usually the right answer.

Now I'm learning to build AI agent systems from the ground up, in public, using that same instinct for workflows, constraints, and what can go wrong. I'm not the expert; I'm just a few steps ahead and showing my work.

Common questions

Do I need to know how to code?

No. I don't come from an engineering background either. The build logs focus on config files, no-code/low-code tools, and understanding how pieces connect, not writing software from scratch.

How much time does this take?

Reading a build log: 10–15 minutes. Following along and building: a few hours per week, at your own pace. There's no curriculum or deadline.

What's the difference between this and AI courses or newsletters?

Most AI content is either too abstract ("AI will change everything") or too advanced (assumes you already code). This is concrete configs, actual failures, and practical wiring, aimed at capable people without engineering backgrounds.

Is this free?

The build log emails and core templates are free. I may eventually offer something paid, but the learning-in-public work will stay accessible.

What tools do you use?

Mostly free or freemium: Claude, ChatGPT, n8n, CrewAI, MCP servers, Google Sheets/Docs, Notion. The goal is simple: accessible stacks and money left for LLM APIs, not pricey enterprise platforms.

Follow the build

Get weekly(ish) build logs with real configs, failures, and fixes. No spam. Unsubscribe anytime.