/EdgeLog ยท Essays

Essays

28 pieces, 3 axes, 5 series. Every claim carries a number.

Series NeuroTerm 2 published
01 Why I Started Building NeuroTerm 4 min · 02 How I Keep Speed, Quality and Intent with AI Agents 8 min ·
Series harness-maker 2 published
01 Side vs Production 4 min · 02 Intent-Based Development 8 min ·
Series Tuned 5 published
01 What I Am Making, and Why I Make It Alone 3 min · 02 The Co-Founder That Built Too Fast 4 min · 03 A Roadmap That Marks Where to Quit First 5 min · 04 The Two Seats Left to People 6 min · 05 What I Don't Hand Off to My AI Team 5 min ·
Series AI-Written Firmware 4 published · 1 upcoming
01 What I Didn't Tell It 5 min · 02 What Breaks Outside the Function 8 min · 03 Measuring the Gap With 233 Tests 5 min · 04 What to Hand Over, Where to Stop It 7 min ·
05 Introducing the Verification Gate Releases
Series AI Agents in Embedded 3 published
01 The Spec-Driven TDD We Put Off 9 min · 02 Learning Context Engineering 9 min · 03 Opening the Full Pipeline 10 min ·

11 standalone essays

  1. 01 Philosophy Making What I Don't Fully Understand I don't fully understand the products I make. We always used things we couldn't see into, for reasons we trusted. As those reasons fade, what remains? · 5 min ·
  2. 02 Philosophy I Don't Know My Philosophy Yet I spent years building whatever the requirements said. Making my own products, I found I needed a philosophy. I have none yet, so I write to find it. · 2 min ·
  3. 03 Methods How Many AI Code Reviews Is Enough? I spent 400 LLM calls on one codebase to learn this. One review pass finds 34% of the real defects, and the four most-repeated findings were all false. · 30 min ·
  4. 04 Methods LLM PCB Routing: What I Learned From a Human's Fixes I handed a 96% auto-routed PCB to a contract engineer. He redrew finished routes, not only the last 4%. I kept that difference as 16 rules for my agent. · 16 min ·
  5. 05 Methods How Long Does Harness Engineering Have? Everything useful in a harness ends up absorbed into vanilla. What survives is what the vendor structurally will not or cannot build. Notes on the shift. · 2 min ·
  6. 06 Methods What loop would keep my software improving with LLMs? A hypothesis I am still sorting out: whether I can wire my own project to LLMs and turn it into a loop that keeps improving itself. · 10 min ·
  7. 07 Methods The CRA Begins: Embedded Software's Center of Gravity Moves September 11, 2026: the EU Cyber Resilience Act's first duty applies. I see it less as compliance than as embedded engineering's center of gravity moving. · 8 min ·
  8. 08 Methods AI Finally Needs Hardware Engineers In 2026 inference overtook training. The edge is where it grows fastest, and the work left there, I think, is the work AI cannot do. · 5 min ·
  9. 09 Methods Offline Datasheet RAG: The Pipeline I Argued Against I argued against RAG for hardware docs, then an offline job made me build one. What survived: SVD parsing, structural chunks, hybrid search, multi-query. · 6 min ·
  10. 10 Methods Hardware Docs for LLMs: Why I Went Back to grep RAG cut the cross-references that make hardware docs useful. After wrestling with a pipeline, I went back to handing the LLM my SVD and PDF as file tools. · 9 min ·
  11. 11 Making Why I Started EdgeLog I wanted to know what happens when LLMs, zero-trust security, and modern frameworks meet real embedded hardware. A research log from the edge. · 4 min ·