Article 5: Track everything
Article five of my personal coding constitution: every action that mutates state must leave evidence. The trail isn't for the auditors. It's the thing that earns me the right to keep going.
The atomic-unit architecture, twelve months in
A year ago I wrote about treating each AI interaction as its own bounded unit, own context, own audit, own memory boundary. Twelve months of building inside that pattern is enough material to grade what held up, what didn't, and what I missed entirely.
AI in the news: week of March 8, 2026
MWC Barcelona puts agentic AI in every booth, GPT-5.4 ships mid-week, Oracle telegraphs the largest layoffs in its history to fund AI capex, and the EU Council adds non-consensual deepfakes to the AI Act's prohibited list. The week the cuts started being about funding compute, not replacing workers.
Notes on building your own AI assistant: start here
Starter notes for the practitioner who wants to build their own personal AI in 2026. The substrate is finally ready. Hardware, model, MCP-or-not, memory, privacy boundaries, the pragmatic shape of 'if I were starting today.'
Why MCP + everything else for tool integration in 2026
MCP won the integration layer. That doesn't mean every tool integration in 2026 should be MCP. The honest architecture is MCP for 80% of cases plus a small set of deliberate escape hatches for the 20% where the protocol shape doesn't fit. Worth being specific.
The model that works on your laptop: March 2026 edition
The recurring laptop-and-Studio model rundown, March 2026 edition. Llama 4, Qwen 3, Mistral 3, Kimi K2 Thinking, and the small distills, what runs where, what it's actually good for, what I'd reach for first.
AI in the news: week of March 1, 2026
MWC Barcelona pushes on-device AI back to the front of the stack, AMD and Nutanix sign a multi-year agentic-AI deal that makes the on-prem enterprise stack real, HyperNova 60B drops on Hugging Face, and the state-AI-law map keeps moving while the federal preemption clock ticks.
When AI helps and when it doesn't
A year of using AI in daily work. Concrete cases where it earned its keep, concrete cases where it didn't, and the pattern that explains the difference. Worth being honest about both halves of the practitioner experience.
The single-user → multi-tenant migration I actually shipped
The playbook for a single-tenant to multi-tenant migration done well, additive schema, identity backfill, dual-path policy enforcement, row-level security, observability for both paths, traffic shift, deprecation. What I wish I'd known on day 1.
Privacy by design for personal AI
Privacy-by-design for personal AI in 2026 isn't a policy posture, it's an architecture. Local-first compute, redaction layers, scoped access, audit trails, deliberate sync. The patterns the principled-user community has converged on, written down concretely.
AI in the news: week of February 22, 2026
Gemini 3.1 Pro and Sonnet 4.6 land inside a week, Alibaba slips Qwen3.5 out hours before Lunar New Year, the EU stretches its AI Act timeline, and Baker McKenzie cuts 600-1,000 attorneys citing AI. The cadence is compressed; white-collar is now in scope.
Governance for a team of one
Enterprise AI governance frameworks don't scale down to a solo operator, and yet governance is one of the four positions I keep coming back to. Here's the lightweight-but-real version I actually run, and why each piece earns its keep when nobody else is going to ask.