#6048: Under the Hood, October 2026: Seventeen Prompt Versions and Three-Times-Faster Voices

Four weeks of pipeline changes, grounded in the code: grace periods, duplicate detection, Hilbert's rules, echo deletion, and voices nearly three times faster.

workflow-automationai-inferenceprompt-engineering

#6047: A Human Guest on His Own AI Podcast

Daniel built an AI podcast with cloned voices — then sat down as a guest on it. What six thousand episodes taught him about prompts.

ai-coding-assistantsprompt-engineeringspeech-recognition

#6046: Five Jobs, One Laptop: The Overemployed Playbook

A British civil servant allegedly held five full-time jobs at once — until his make-up started separating on a Teams call. What the overemployed pl...

workplace-culturecareers-and-hiringfraud-and-scams

#6045: The Two-Week Job You Never Explain

A White House press secretary lasted five days. What the roll call of spectacular flameouts says about the two-week gap on your CV.

careers-and-hiringworkplace-cultureworkplace-culture

#6044: One Door for Every AI: Cloudflare MCP Portals

Cloudflare's MCP Portals put every on-prem MCP server behind one authenticated endpoint — and why service tokens can't work for ChatGPT.

model-context-protocolcyberattacksauthentication

#6043: Write the Spec Before You Pick the Model

Long context, character voice, cheap at scale — three requirements that fight each other. The document that decides the fight.

ai-evaluationprompt-engineeringhallucinations

#6042: Who Grades the Grader? AI Writing Benchmarks Put to the Test

Maths benchmarks can't see whether a model's prose collapses by chapter three. Five specialist leaderboards try — and our own writer is on them.

ai-evaluationai-biasprompt-engineering

#6041: How CENO Moves a Web Page Into a Censored Country

A walk through CENO's actual plumbing: signed caches, volunteer bridges, and why the chain is deliberately short.

internet-infrastructuredigital-privacycryptography

#6040: Claude Code Subagents Can Now Spawn Subagents

Claude Code subagents can now spawn their own subagents. Here's where the depth limits come from, and why nobody actually knows where they stop.

multi-agent-systemsai-coding-assistantscontext-and-memory

#6039: Who's Actually Building Frontier AI Outside the US and China?

Mistral sits 20 points off the frontier. Korea runs a bake-off. The UAE open-sources everything. What's actually driving sovereign AI?

ai-evaluationopen-sourcefine-tuning

#6038: Chinese Frontier Models, Cost vs Capability

Eight Chinese labs, two tiers each, and a pricing spreadsheet that says your default model is mid-pack.

open-sourceai-inferencelocal-ai

#6037: What's New in Gemini 4 Argon

A close read of Gemini 4 Argon's model card — a million-token window, thinking levels, implicit-only caching, and a lot of blanks.

ai-evaluationai-inferencecontext-and-memory

#6036: GPT-6 Astra: A Million Tokens, No Fine-Tuning

A million-token window, ten built-in tools, a 272K pricing cliff — and no fine-tuning. What GPT-6 Astra actually offers builders.

ai-evaluationai-inferencecontext-and-memory

#6035: Who Else Is Making AI Podcasts? A Survey

A survey of AI-hosted podcasts: one full match, a 730-episode run, a home NAS rig, and a graveyard of good tooling.

ai-coding-assistantsmulti-agent-systemsopen-source

#6034: Claude Opus 5.5: What Anthropic Actually Shipped

Anthropic's new flagship is cheaper than the model it replaces — and won't let you turn thinking off.

ai-inferenceai-evaluationlocal-ai

#6033: Progressive Skills Discovery: When Bigger Context Windows Aren't Enough

Bigger context windows were supposed to make skill discovery moot. Here's why the listing budget still bites at scale.

context-and-memorymodel-context-protocolai-coding-assistants

#6032: Blending Datasets vs Merging Weights for Fine-Tuning

Merging weights happens in parameter space. Blending datasets happens in data space — and the messy part is worth it.

fine-tuningtraining-dataai-evaluation

#6031: Why Two Parents Alone Can't Carry a Toddler

Daycare didn't fix it. The research says the problem isn't hours — it's fragmented hours, portable mental load, and a village that isn't there.

childcare-and-daycareworkflow-automationproductivity

#6030: Who Do You Actually Hire to Train a Model?

Research engineers and data engineers are diverging. Here's who you actually hire to train a model — and why the org chart keeps changing.

training-dataworkplace-culturefuture-of-work

#6029: One Model, Two Modes: Does AI Consolidation Hold Up?

OpenAI hides a router, DeepSeek merged two modes, Anthropic removed the knob. Is the model landscape really consolidating?

transformer-architectureai-inferencefine-tuning

#6028: ASR vs Multimodal LLMs: What Actually Won

Dedicated ASR plus a cleanup model is quietly beating end-to-end dictation on the hardware you already own.

speech-recognitiontext-to-speechai-evaluation

#6027: Can You Snap Two Fine-Tunes Together Like LEGO?

Two fine-tunes, one base model, and a question from Jerusalem: can you merge Hebrew code-switching with technical vocab — without retraining?

speech-recognitionfine-tuninglocal-ai

#6025: When Should Claude Code Start a New Session?

Long context windows don't fix stale context — they hide it. Why task boundaries, not token counts, are the real problem.

ai-coding-assistantscontext-and-memorymulti-agent-systems

#6024: What's Actually Inside the KV Cache?

890 bytes per token: why the KV cache is both what makes LLMs practical and what makes long context expensive.

ai-inferencecontext-and-memorygpus

#6023: How Much Voice Data Is Enough?

Two hours of phone recordings took Hebrew code-switching from 20% to 60%. But when does more data stop mattering?

speech-recognitionfine-tuningtraining-data

#6022: Fine-tuning ASR for code-switched Hebrew

Two hours of one man's voice beat models trained on 120,000 hours — but only for code-switched Hebrew.

speech-recognitionfine-tuningtraining-data

#6021: Why Chat Templates Break Models Silently

Every chat API flattens your messages into special tokens. Get one delimiter wrong and the model just gets quietly dumber.

transformer-architectureprompt-engineeringai-inference

#6020: What Buildings Do to Your Body

Measured cortisol, pupil dilation, and fractal dimensions reveal how traditional and modernist buildings shape your nervous system.

stress-and-traumacognitive-biasessleep-and-circadian-rhythm

#6019: Who Should Decide What Gets Built?

A listener asks what developers' role in a city should really be — and two very different cities show what happens when you get the answer right or...

zoninghousing-affordabilityurban-design

#6018: The Plan Nobody Has Ever Read

Jerusalem's Taba is a legal document no single person has ever read cover to cover — and it governs where millions of people can build.

zoninghousing-affordabilityisrael

#6017: Israel's 319-Day Permit: Why Building Takes a Year

Israel averages 319 days to a building permit — nearly 3x the Western average. Where does the year actually go?

zoninghousing-affordabilityconstruction