Mark & Carl talk with Swizec Teller about using AI at work

Transcript ★ Support this podcast ★ Reply on Bluesky (00:00) - This Month in React April (00:26) - Introductions (02:36) - What convinced you AI code tools were worth using? (04:24) - Using early ChatGPT for DB migrations (06:59) - Watching AI use a command-line (08:16) - Background chat agents (08:58) - Staying very hands-on while using AI tools (11:00) - Driving AI closely without reading its code (17:50) - Mark's workflow; OpenCode with CodeNomad UI, plus IDE+git UI. Opus 4.6 on API (20:39) - Swizec's workflow, latest Cursor on Opus (23:58) - Carl's workflow, mostly ClaudeCode but looking at custom orchestrators (25:23) - Exploring fully autonomous agents (28:08) - Mark's AI debugging work in React core (31:42) - Value of providing more context (33:58) - AI-owned documentation (37:09) - Using good engineering practices still matters? (40:47) - How do you know the right code to make? (42:49) - Good communication still matters (45:13) - What will "review" look like in the future? (46:17) - Automating functionality tests with deployment practices (51:49) - What behaviors belong to the agent, and what fundamentally can't be part of the agent? (56:17) - Impacts of LLMs on software engineering? (01:03:22) - A superpower right now is a domain expert who can kind of code