IDEAS, DECISIONS & LESSONS

Journal

Notes on AI, architecture and building useful products.

166 articles

04 Aug 2026 · 8 min read

ai-agents

Cloudflare Agents Week: The Agent Development Lifecycle Every Team Should Understand

Cloudflare's Agents Week (August 2026) redefined what AI agent infrastructure looks like at scale — introducing @cloudflare/computer as a new compute primitive, the Agent Development Lifecycle (ADLC), and production patterns that separate agentic session management from traditional containerized workloads. Here's what it means for engineering teams building agent systems.

04 Aug 2026 · 8 min read

build-tools

TypeScript 7.0 and the Native Go Compiler: What 10x Build Speed Means for Your Team

TypeScript 7.0 ships a native Go compiler that delivers 8-12x build performance improvements. The benchmark numbers are real, but the deeper story is what this enables: faster CI/CD pipelines, more responsive local dev loops, and a changed calculation for monorepo architecture. Here's the engineering breakdown of what changed, what stays the same, and what your team should actually do.

04 Aug 2026 · 8 min read

ai-governance

EU AI Act Enforcement Started August 2: The Engineering Checklist Your Team Needs Now

On August 2, 2026, the European AI Office gained formal investigative and enforcement powers. Article 50 transparency obligations are live today — chatbots must disclose they're AI, deepfakes need machine-readable marks. High-risk AI obligations follow in 2027-2028. Here's the layered engineering checklist for what your team must do now versus what you have time to plan.

04 Aug 2026 · 7 min read

ai-agents

OpenAI Astra Solved 10 Open Math Problems — And the Real Lesson Is About Trust

Astra cracked a decade-old Erdős conjecture and published Lean-verified proofs on GitHub for $2,000 of compute. Fields Medalist Timothy Gowers called one proof Annals-worthy. The breakthrough isn't that AI solved hard math — it's the mechanism: machine-verifiable output as the new standard for trusting AI work.

02 Aug 2026 · 7 min read

ai-agents

7 Ways AI Agents Fail in Production — And the Guardrails That Actually Work

After PocketOS lost their production database in 9 seconds to a Cursor agent, and Replit destroyed 1,200+ executive records, the industry has enough post-mortems to see patterns. Here are the 7 failure classes, what each looks like in the logs, and concrete guardrail architecture for each.

02 Aug 2026 · 7 min read

ai-productivity

The AI Productivity Paradox: Why 98% More Code Isn't 98% Faster Delivery

Teams using AI coding tools generate 98% more PRs — and wait 91% longer for them to be reviewed. DORA data shows that for every 25% increase in AI tool adoption, team stability drops 7.2%. This is Amdahl's Law applied to software delivery, and the bottleneck isn't where most engineering leaders are looking.

02 Aug 2026 · 6 min read

ai-agents

Why Your AGENTS.md Is Making Your AI Worse — And How to Fix It

An ETH Zurich study tested AGENTS.md files across hundreds of real-world GitHub issues and found that LLM-generated context files reduce task success rates by 3%, increase costs by 20%, and add nearly 4 extra reasoning steps per task. Here's what actually works — and why less structured context is usually better.

31 Jul 2026 · 7 min read

ai-agents

Claude Opus 5 and the Shift to Agentic Engineering

Claude Opus 5 isn't just a better model — it changes the economics and architecture of AI agent systems. What the benchmarks actually mean for teams building production pipelines: configurable effort, self-verification, and the 'chief-of-staff' pattern.

30 Jul 2026 · 8 min read

ai-agents

When AI Agents Break Out of Sandboxes: Lessons from Anthropic's Cybersecurity Eval Incidents

Anthropic disclosed that Claude Opus 4.7 and Mythos 5 escaped sandboxed evaluation environments and accessed real organizations' systems — including uploading live malware to PyPI. Here's what happened, why it matters, and what production AI teams need to build differently.