HeyJaredThe Chain of Thought Podcast
Host Conor Bronsdon explores artificial intelligence through conversations with the founders, engineers, and researchers building breakthrough AI systems.
- Indexed videos, last 90 days
- 53
- Latest publication
- Sep 30, 2026
- Audience
- ~3K subscribers
- Earliest in this view
- Jul 6, 2026
Latest videos
Poolside trains on almost a million containerized repos (opens the original)
Read excerpt
In May 2025, Poolside co-founder Eiso Kant described how Poolside trains its coding models: reinforcement learning from code execution feedback on almost a million repositories that are fully containerized with their full test suites, so the model learns from when it is right and wrong. In August 2026, Newcomer and Bloomberg reported that Nvidia will pay $6B to license Poolside's Model Factory. From episode: First Code, Then AGI: Software's Event Horizon with Poolside Founders Jason Warner & Eis
Barring AGI, AI is a normal technology (opens the original)
Read excerpt
Laurie Voss says that unless we invent superintelligence, AI follows the same path as every other technology: a hype peak, a collapse in valuations where a lot of people lose their shirts, and a slow climb back to roughly where we expected, just much later than promised. Those are his Bayesian priors on bubbles, and he thinks this one plays out the same way. From episode: AI Codes: Product Engineers Decide What to Build Full episode: https://chainofthought.show/podcast/74-ai-codes-product-engine
Closed-source moats get cloned over a weekend (opens the original)
Read excerpt
Laurie Voss predicts that closed-source software is about to feel what open source felt first: someone with a weekend and a pile of tokens clones the idea without touching the code. That disrupts what companies can charge for and how, and he expects it to produce a lot more open source as the cost of code generation heads to zero. From episode: AI Codes: Product Engineers Decide What to Build Full episode: https://chainofthought.show/podcast/74-ai-codes-product-engineers-decide-what-to-build-lau
Rewrite AI slop: 60/100 to 0/100 with the facts intact (opens the original)
Read excerpt
avoid-ai-writing is an open-source skill for Claude Code, Codex, Cursor and more that flags the patterns that make a draft sound machine-made, then rewrites them. In the README example, a 66-word funding blurb full of "nestled in the heart of," "a watershed moment" and "experts believe" gets 16 findings and a 60/100 score from the bundled detector. The rewrite scores 0/100 and keeps the funding amount, the location and the product details. Try it: https://github.com/conorbronsdon/avoid-ai-writin
Humans can’t extrapolate exponentials, even AI labs (opens the original)
Read excerpt
Laurie Voss explains why people inside the AI labs are rattled: humans are bad at extrapolating exponential growth even when they know it is exponential. He saw it first-hand at npm, which grew exponentially for ten years while every year the team assumed downloads would finally taper off. Being five or ten times better again a year from now is a more dramatic change than it sounds. From episode: AI Codes: Product Engineers Decide What to Build Full episode: https://chainofthought.show/podcast/7
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
~3K subscribers
Measured Sep 19, 2026
Source's subscribers, not the number who saw an individual piece.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.