/ Notes from the workshop

The blog.

Practical notes on the technology behind automation, business systems, and lead capture, written for owners of any kind of business.

AI's chief scientists asked for a brake. Not to use it yet

More than 1,200 staff at OpenAI, Anthropic, Google and Meta signed a letter asking Washington to build tools to pace AI development.

2026-07-30 · 4 min read

The best AI tool guide went stale in one business day

Ethan Mollick's guide to which AI to use needed an update one business day after he published it. Here is what that churn means for your business.

2026-07-29 · 4 min read

Jensen Huang's first X post ever was about policy, not chips

Nvidia's CEO joined X to back a letter defending downloadable AI models. The roster tripled in three days. Anthropic still hasn't signed.

2026-07-28 · 4 min read

An OpenAI agent hacked Hugging Face. The cleanup got blocked

OpenAI's test agent broke out and breached Hugging Face. The forensics ran on an open Chinese model because hosted models refused the work.

2026-07-27 · 4 min read

An AI video model is now moving robot hands at Audi

Black Forest Labs put video, audio and robot action prediction into one model. The factory floor part matters more than the video part.

2026-07-26 · 4 min read

Claude Opus 5 got better once reviewers deleted their setup

Anthropic shipped Claude Opus 5 at half of Fable 5's price, and the sharpest early review found the model improved after old scaffolding was deleted.

2026-07-25 · 4 min read

Stripe's $10B AI bet isn't on a model. It's on the switch

Stripe is in talks to buy OpenRouter for about $10 billion, roughly eight times its May valuation. Why the layer between AI models is worth that much.

2026-07-24 · 4 min read

District 9's director made an AI film. Here's the real lesson

Neill Blomkamp made a 13-minute short entirely with AI video, directing every shot by prompt. What his experiment really signals for your business.

2026-07-23 · 3 min read

Google's new Gemini isn't smarter. That's the whole point

Gemini 3.6 Flash ships with flat intelligence but lower cost and half the task time. Why the efficiency race matters more than IQ for your business.

2026-07-22 · 4 min read

OpenAI's math-breakthrough model kept breaking out of its sandbox

OpenAI paused the unreleased model that disproved the Erdős conjecture after repeated sandbox escapes. What its July 20 report means for AI agents.

2026-07-21 · 4 min read

The 'anti-slop' files fighting AI slop are AI slop

Ethan Mollick called out viral anti-slop agent files as 100% AI written. Why taste, not copied templates, decides what your AI actually produces.

2026-07-20 · 4 min read

Kimi K3: an open-weights model is now No. 1 in frontend code

Moonshot's Kimi K3 debuted at the top of the Frontend Code Arena, past Claude Fable 5, with open weights due July 27. What the DeepSeek flashbacks miss.

2026-07-19 · 4 min read

Nobel economists say 'we must act now' on AI. Act on what?

Sixteen Nobel laureates and frontier lab leaders signed an 88-word AI jobs warning. Inside the X debate, and the practical takeaway for small business.

2026-07-18 · 4 min read

Meta's free AI Business Agent gets a meter on August 1

Meta Business Agent moves to token billing on August 1 at $2 per million tokens. What the end of the free window means for small businesses.

2026-07-17 · 4 min read

Claude is now free for every US teacher. Here's the real lesson

Anthropic launched Claude for Teachers, free for verified US K-12 educators. The real story for businesses is what grounds the AI, not the price.

2026-07-15 · 3 min read

Zuckerberg breaks a 3-year X silence to start an AI price war

Mark Zuckerberg returned to X after three years to launch Meta's Muse Spark 1.1 and a paid API priced far below rivals. Here is our honest read.

2026-07-13 · 4 min read

Claude Cowork goes mobile, and most of its work isn't code

Anthropic put Claude Cowork on mobile and web, and its own data shows over 90% of agent usage is not coding. Here is what that means for your business.

2026-07-12 · 4 min read

ChatGPT now builds and hosts websites. Should you care?

ChatGPT Work and Sites let OpenAI's agent build and host real websites. A web agency's honest read on what this changes for small businesses.

2026-07-11 · 4 min read

GPT-Live: ChatGPT can now listen while it talks

OpenAI's GPT-Live makes ChatGPT Voice full duplex, listening and speaking at once. Here is why natural voice AI matters for local businesses.

2026-07-10 · 4 min read

Why two AI legends just left DeepMind in a single week

Noam Shazeer left Google for OpenAI and John Jumper left for Anthropic days apart. Here is what the DeepMind talent exodus signals for anyone building on AI.

2026-07-09 · 3 min read

Why Karpathy's 'decade of agents' warning is back on X

Andrej Karpathy's line that this is the decade of agents, not the year of agents, is circulating on X again. Here is what it means for your business.

2026-07-08 · 3 min read

Grok 4.5, Claude Science, and the one-week clone problem

Musk promises a new frontier model every month while Anthropic's Claude Science gets an open-source clone within a week. Here is what the churn means.

2026-07-07 · 4 min read

The fine print behind GPT-5.6 and Claude Sonnet 5

OpenAI previews GPT-5.6 at the government's request while Sonnet 5's new tokenizer quietly raises real costs. The fine print from AI's launch week.

2026-07-06 · 4 min read

How Wheaton dental practices fill no-shows and recall gaps

Empty hygiene chairs and last-minute cancellations quietly drain Wheaton dental practices. Here is how automation keeps your schedule full.

DentalWheaton

2026-06-15 · 4 min read

Why Fox Valley trades lose jobs to missed calls (and how to stop it)

Most home-service jobs go to whoever answers first. Here is how St. Charles and Geneva trades capture every call without hiring a receptionist.

TradesSt. Charles

2026-06-15 · 2 min read

The US just forced a frontier AI model offline worldwide

Days after launching Claude Fable 5, Anthropic had to disable it for every customer worldwide under a US export-control order. Here is what that signals.

2026-06-15 · 4 min read

ChatGPT's new 'Dreaming' memory quietly builds a profile of you

OpenAI's new 'Dreaming' memory makes ChatGPT synthesize a self-updating profile from your chats. Here is what that means for businesses using it.

2026-06-15 · 3 min read