25 October 2026|AI Engineering
454 Pull Requests in Twelve Days
A month of unattended building on a live product: 454 pull requests in twelve days, six production releases and no reverts, and the pieces that made it safe.
Engineering
Long-form write-ups of how I build with agents, with the numbers. Memory, model routing, review gates and the hardware underneath.
6 September 2026|Agent Grounding
What the Agent Does Not Know to Look For
The standard answer to “how do I stop my agent making things up” is retrieval. Give it a search tool over your documents and your code, and it will look…
30 July 2026|AI Engineering
The Cheapest Model That Can Safely Do the Job
How do you start using bigger models for the complex tasks without driving the cost through the roof? My last article described a delivery pipeline end to…
28 July 2026|AI Engineering
The Drawing Office and the Yard
In my last article I described what a closed build loop needs – a foreman that owns a backlog, a build that runs a real correctness loop, independent review…
17 July 2026|AI Engineering
Closing the Loop to Full Autonomy
The most important button in my development process now does almost nothing. When a pull request lands for review on one of my products, I read the summary…
14 July 2026|AI Engineering
Shipping While We Sleep
One evening in June I pointed an AI agent at a product backlog, gave it the rules of engagement, and went to bed.
26 May 2026|Agent Grounding
Resumption-First Memory: Design Rules for Agentic Systems
Every serious agentic memory system you can read about today has converged on roughly the same whiteboard. Capture the turns. Summarise them.
22 May 2026|Agent Grounding
Surviving Claude Compaction: How to Make Your Sessions Outlive Their Context Window
Compaction wipes what an agent session has learned. Five tiers of session memory, from files Claude Code already writes to a shared team brain.
9 May 2026|AI Engineering
Diagrams as Mermaid, Scraping With Permission, and Triage That Knows Your Stack
Three upgrades to my second brain: diagrams drawn as Mermaid, scraping that respects a site's terms, and research triage grounded in my own stack.
6 May 2026|Sovereign AI
DGX Spark Optimisation, Four Iterations On: From Defaults to 308 Tokens Per Second Sustained
Four rounds of tuning gpt-oss-120b on one DGX Spark: a 4.3x single-stream gain from flags, an EAGLE3 dead end, a source build and a one-line CUDA graph fix.
1 May 2026|Agent Grounding
The AI Productivity Paradox, and the Team-Context Infrastructure That Fixed Mine
Most developers now use AI tools, yet the measured gains disappoint. Why the paradox happens, and the team-context infrastructure that fixed it for me.
22 April 2026|Sovereign AI
NVIDIA DGX Spark Concurrency Benchmark: 120x Throughput the Single-Stream Reviews Miss
Benchmarking the NVIDIA DGX Spark (GB10, 128GB): 120x more throughput at concurrency 256 than single-stream reviews show, with the full vLLM config.
Nothing matches that yet. Try another word or topic.