Artificial Intelligence
Episodes whose main subject is AI itself: building with, training, evaluating, or debating AI systems. AI used only as a tool for another topic goes to that topic.
#5949: Inside Deep Research Agents: Claims, Citations, and the 84.7% Orchestrator Problem
Why 84.7% of citation errors originate at the orchestrator — and how a claim ledger fixes it.
#5948: LangGraph, LangChain, and Deep Agents in 2026
Runtime, framework, harness — how three LangChain layers actually fit together, and what changed in 2026.
#5946: Inside Humanity's Last Exam: 2,500 Questions AI Can't Ace
2,500 questions, ~1,000 experts, and a top score that jumped from 2.7% to 54.8% in 18 months. What is HLE actually measuring?
#5945: How a Reward Model Actually Trains an LLM
One number, no gradients, four models in the room. What a reward model actually does inside the RLHF loop.
#5944: Keeping the Rollout Engine Fed: Inside RLHF Pipelines
Most of an RLHF run is spent waiting on generation, not gradients. Here's how TRL, OpenRLHF, and verl actually differ.
#5941: Five Minutes Reading Your Dictation Prompt
A listener's speech-to-text cleanup prompt, read out in full — then stress-tested against what ASR actually gets wrong.
#5940: The Missing Middle: Content APIs and ISR for a 6,000-Episode Catalog
When your catalog is too big to rebuild on every change, what has to change? Turns out the missing middle might already exist.
#5939: Android Notifications: PWA vs Native, What You Actually Get
Push, channels, haptics, and the scheduling gap — what a PWA can and can't do that a native Android build can.
#5938: One Regex Rule Worked, One Was Impossible
A dictation app, two text-replacement rules, and the mathematical line between what regex can match and what it provably cannot.
#5937: The Fine-Tuning Career: Titles, Skills, and What It Pays
The job above the labeling layer exists — it just doesn't have one name. What it takes, what it pays, and how to sell it.
#5936: Training a Model When You Don't Know the Labels Yet
Daniel wants to extract adapter specs from his shelf — but he can't name the labels until he's already annotating.
#5934: Why Modiin Is Ugly Despite Its Form-Based Code
Modiin had a form-based code and still came out ugly. The reason sits in what the code refused to regulate: facades, windows, materials.
#5933: Form-Based Codes and the City That Feels Dead
Can a zoning document create a walkable Levantine city with real character — or does over-planning kill the spontaneity that makes cities alive?
#5927: Where Generalists Get Priced: Inside the FDE Role
A $485K median salary for a job with no clean definition — and why the résumé can't capture it.
#5922: Teaching a Model Your Voice: SFT vs DPO
Three ways to teach a model your writing voice — and the evidence says one of them quietly makes things worse.
#5915: How Multi-Agent AI Simulations Keep a Shared World State
Concordia keeps the world in a dictionary an LLM writes to. Generative Agents keeps it in a spreadsheet. CAMEL keeps it in the chat log.
#5913: Why the Sterile Cockpit Rule Took 7 Years
One crash, one morning, one city — and seven years before the FAA wrote the rule down. What Eastern 212 still teaches us about attention.
#5912: Building Your Own CRM with AI: Where to Start
AI can't write your CRM for you — but it can remove the friction. Start with the schema, not the code.
#5910: Revit MCP in a VM: What Actually Works
Autodesk's official Revit MCP server is locked down and reliable-ish. The community ones execute arbitrary code on an open port. Neither runs unatt...
#5908: One Database Won't Fix Your Fragmented Stack
Rita has two employees, four apps, and one fear: which database wins when they disagree? Here's the field-by-field answer.