Last Week in AI: You Can Now Buy Ads Inside ChatGPT
OpenAI launched ChatGPT ads for small businesses and a model it calls AGI-era. For your business, the ads story matters more.
Resources
Practical frameworks and tactical guides on automation, sales systems, and AI tools — written for the people running the business.
OpenAI launched ChatGPT ads for small businesses and a model it calls AGI-era. For your business, the ads story matters more.
Most owners have a file full of numbers they never look at because reading it properly takes an afternoon. It does not anymore.
Ramp data from 70,000 businesses shows Anthropic Fable 5 losing to cheaper models. Pew: 34% use AI for health. Salesforce moved CRM inside Claude.
Copy these five prompts for chasing invoices, raising prices, reading contracts, writing quotes, and talking instead of typing.
A one-click Copilot flaw, Pew finds 52% more concerned than excited about AI, and DeepSeek changed prices twice in eight days.
We sell AI systems for a living. Here is the honest list of what it cannot do — and what it will confidently pretend it can.
Three AI labs disclosed model breaches in three weeks. The same 35-person vendor caused two of them — and nobody had mapped that until it broke.
AI knows your business exists. It just won't recommend you. Also: OpenAI's 80% price cut, Anthropic test breaches, and Nvidia's $250B OpenAI debt talks.
Everyone quotes AI prices. Nobody explains them. Here is what a token is, what $5 per million tokens buys, and what the top models charge right now.
OpenAI models escaped a locked test and hacked Hugging Face. Claude Opus 5 took #1 cheaper, Europe fined Google $1B, and Kimi K3 went open.
Paste this into ChatGPT, Claude, or Gemini. It audits one process, won't flatter you, and hands back the single highest-leverage fix to make first.
Meta opted public Instagram into AI image gen by default, Gemini 3.5 Pro delayed again, and frontier AI got 4–5x cheaper this week.
Your AI tool's price will change again. The businesses that shrug instead of rebuilding ask one different question at the start.
A new frontier model launched this week. Another launched last week. Here is why picking the "best" AI model is the wrong question for a small business.
Autonomous AI ransomware got documented. GPT-5.6 launched, Anthropic passed OpenAI in revenue, and Chinese models now handle nearly half of US enterprise usage.
What an operations assessment actually covers, what you walk away with, and the 15 minutes of prep that makes it twice as useful.
Fable 5 is back with new pricing, AI found a 29-year security bug, HP went all-in on OpenAI, and Amazon renegotiated with Anthropic.
SaaS covers the standard 80% of every business. The 20% that's uniquely yours is where the hours leak. Here's the honest decision framework.
ChatGPT made most businesses faster typists, not automated operators. There's a difference between using AI and running it — and only one compounds.
You're not avoiding AI costs by waiting — you're paying in time. Three hours of daily admin at $75/hr is $4,500/month. Every week you wait compounds.
Most businesses have the tools. Almost none have the layer that connects them. Here's the four-layer framework that separates manual ops from automated ops.
Audience
Domain
Type