IDEAS, DECISIONS & LESSONS
Notes on AI, architecture and building useful products.
279 articles
21 Jul 2026 · 4 min read
ai
Moonshot AI's Kimi K3 just beat Claude Fable 5 on Frontend Code Arena. Here's how to evaluate any new open-weight model release as a Tech Lead — from benchmarks to real production tasks.
Read more21 Jul 2026 · 6 min read
ai
Devin, GitHub Copilot Workspace, Augment Code — AI coding agents are shifting from individual dev tools to shared team infrastructure. Here's what Tech Leads must do now to stay ahead.
Read more20 Jul 2026 · 7 min read
agentic-ai
Google's ARD specification (v0.9, 2026) defines how AI agents discover tools at runtime via ai-catalog.json. A Tech Lead's practical guide: what it means, how it compares to MCP, and how to make your platform agent-discoverable today.
Read more19 Jul 2026 · 6 min read
ai
When AI writes the code, reviewing the diff is already too late. Learn the upstream review model — reviewing intent, architecture, and test contracts before a single line is generated.
Read more18 Jul 2026 · 8 min read
agentic-ai
A practical Tech Lead guide to prompt injection attacks, jailbreaking, and defense patterns — with real 2026 incidents, code examples, and actionable mitigations for production AI agents.
Read more17 Jul 2026 · 6 min read
ai
When to fine-tune vs. prompt-engineer, which technique to pick in 2026 (LoRA, QLoRA, DPO), how to size your dataset, and how to wire the whole thing into a CI/CD quality gate that blocks regressions before they ship.
Read more16 Jul 2026 · 7 min read
ai
GPT-5.6, Claude Sonnet 5, and Grok 4.5 all launched within 10 days of each other in July 2026. No single model wins every workload. Here's how to build a routing layer that cuts your AI bill by 40-80% without sacrificing quality.
Read more27 Jun 2026 · 7 min read
ai
JetBrains surveyed 11,000 developers — 90% use AI at work but can't measure ROI. This is the cloud adoption problem repeating itself. Here's the governance framework that actually works.
Read more27 Jun 2026 · 6 min read
ai
GLM-5.2 hits 81.0 on Terminal-Bench and 62.1 on SWE-bench Pro with an MIT license and 1M token context. The gap between open and closed models is closing fast — here's how tech leads should respond.
Read more
