Artificial Developer Intelligence
Three engineer friends argue about AI so you don't have to. Model releases, research papers, tools, and the occasional existential crisis — no LinkedIn cringe required.
Latest Episodes
Ep 41 Oct 2, 2026Meta Muse's 0-Day, DHH's Rails World Keynote, GPT-6 Astra & the LLMentalist Effect
A military chatbot flagged a passing ship as Chinese and possibly nuclear; the boarding was called off at the last minute. The full trio returns from a two-week break for Meta Muse's endpoint-swapping 0-day, DHH's "pencils down" Rails World keynote, a news lightning round, a GPT-6 Astra review and its looped-transformer monitorability worry, the plunging price of thought, telling agents the why, the LLMentalist effect, and a bubble clock that holds at 4:15.
Ep 40 Sep 11, 2026The Psychology of Software Teams with Dr. Cat Hicks
Engineering teams that lowered production pressure to make space for learning ended up roughly 18 months ahead of everyone else. No news, no clock: Shimin and Dan sit down with Dr. Cat Hicks — psychologist for the humans in tech, principal scientist at Catharsis, and author of The Psychology of Software Teams — on brains-in-jars organizations, the 10x engineer myth, the learning-opportunities skill, the fluency illusion, and what a healthy AI-native team actually looks like.
Ep 39 Sep 4, 2026Meta's AI Backfires, GLM 5.3 Flash, OpenAI's Jalapeño Chip & the End of Programming
One developer used Fable V to rewrite Bun in Rust — 1M+ lines, ~7,000 commits, 11 days, ~$165K in API costs — and it now underpins Claude Code. The full trio returns for Meta's backfired AI-native reorg, OpenAI cutting Cursor off after the SpaceX acquisition, the debut of the Model Review segment with ZAI's GLM 5.3 Flash, OpenAI's Jalapeño inference chip, and an AI bubble clock that moves backward for the first time in a month.
Ep 38 Aug 28, 2026AI Homework Atrophy, GitHub Commits Double, Nick Muy Sit-Down & AI Sandbagging
We gave four AI models the same question from three user profiles — all four gave worse answers to the user they judged unable to check them. Shimin and Dan sit down with guest co-host Nick Muy (CISO & VP of Platform Engineering, strut.io) for the 26,000-student AI-homework study, GitHub's commits doubling in four months, why orchestrating agents made every developer a middle manager, and Stripe buying OpenRouter because "the singularity started in January."
Ep 37 Aug 22, 2026Claude Watermarks, Zuck's Superintelligence Essay, Zed's ZDB & Multi-Agent Turf Wars
Anthropic put AI agents on one computer with conflicting goals — they wrote self-replicating malware, killed each other's processes in a loop, and revoked each other's sudo access. Shimin and Dan (Rahul away on vacation number two) cover Claude's new imperceptible watermarks and the one-prompt job of removing them, Zuck's 6,500-word superintelligence essay and its unstated ad model, Zed's ZDB — edit-level deltas, CRDT worktrees, and the LLM conversation stored with the change — Noema on why everyone in tech is so sad, and Anthropic's emergent multi-agent systems research, where a coordinated swarm found 266 vulnerabilities to parallel agents' 21 and newer models negotiate truces instead of sabotaging each other.