
I was on the road this week speaking at a conference, so instead of a stack of daily issues, here is one weekly summary with the best of what you missed. Starting with the big one.
Claude Fable 5.2 just launched, and it costs less than the model it replaces. Anthropic released Fable 5.2 on Monday, November 9, and it's open to everyone, not just big accounts. Pricing dropped about 20%: input went from $10 to $8 per million tokens, and output from $50 to $40. (Tokens are the chunks of text an AI reads and writes, roughly three-quarters of a word each, and you pay by the million.) Anthropic says answers come back up to 30% faster than Fable 5.1, and there's a new fast mode that's up to 2.5 times quicker at a higher price ($16 input, $80 output). The company also reports about 55% fewer blocked requests on harmless questions in early testing, which is the "I can't help with that" nonsense that wastes your afternoon. A sister model, Mythos 5.2, has looser safeguards and is limited to vetted organizations, so you won't be using it. It runs on Anthropic's own platform plus AWS, Google Cloud, and Microsoft Azure. All the performance numbers come from Anthropic itself, so treat them as claims until you test it on your own work. If you use Claude through the app, you don't need to do anything. If you pay for it through the API, your bill just went down. Full launch details at The Next Web.
Claude is now attending your meetings. Anthropic launched Claude Meetings in beta for Team and Enterprise plans. It joins your Google Meet calls as a visible participant, then drafts action plans, maps timelines on a whiteboard, and pulls data from the work apps you've connected. Claude Sheets builds shared spreadsheets with formulas and dropdowns, and Claude Data, which turns your CRM or data warehouse into dashboards you can actually read, is now fully available. If you already pay for Claude, that's a real cut in the after-meeting admin pile. More at TechCrunch.
ChatGPT Ads changed how bidding works. OpenAI is retiring Max Bid and Max CPC and replacing them with Target CPA and Target CPC. With Target CPA, you tell the system what you want to pay on average for each conversion (a sale, a form fill, whatever you define), and it adjusts individual bids to hit that average. Old campaigns keep running until January 1, 2027, then migrate automatically. Two things to do now: make sure your Pixel or Conversions API is actually firing (that's the small piece of code that tells the ad platform a purchase happened), and set targets based on what you've really been paying, not what you wish you were paying. Set them too tight and your ads just show less. Details from OpenAI.
Your buyers are researching you with AI, and you should look at what it says. OpenAI's consumer product lead said ChatGPT now handles roughly 20% of Google's search volume, with 1.2 billion weekly users, up 200 million in two months. Shelly Palmer's point is the uncomfortable follow-up: your customers are already using AI assistants to compare vendors and decide who to trust, and if you haven't used one yourself, you don't know how your business looks through that lens. Go ask one about your company this week. Palmer's piece is worth five minutes.
Cursor can now build and deploy an internal app in one click. The feature is called Cloud Agents. Describe what you want in plain English, say a dashboard of open invoices by client, and it builds the app and sets up hosting, a database, and login. No backend developer. One catch: Cursor's hosting is "as is" with no uptime guarantee, so check the data terms before you put anything client-facing or sensitive on it. See how it works.
Here's your reality check on AI agents. You've heard agents will book travel, place orders, and run customer service. In Uber's busiest week of October, AI agents completed 4 rides and 500 food orders, out of roughly 320 million total. The CEO called it "early days," which is the polite way to say almost nothing. The technology is real. Mass adoption is not, so don't rebuild your customer workflow around it yet. The Verge has the numbers.
Two-thirds of small businesses say they use AI. About one in seven has built it into the work that matters. Census and Goldman Sachs data put adoption at 66%, but only around 15% have woven it into core operations. The rest prompt occasionally and call it a strategy. If you feel like you're doing AI but nothing has gotten faster, that's your diagnosis: there's a big difference between using a tool and changing a workflow. Forbes breaks down why.
Also Worth Knowing
A developer built a full AI video editor in three days with Fable 5.2: the trick was handing the model the complete product brief upfront instead of feeding it steps one at a time. Front-load the context, then get out of the way.
Anthropic's amended IPO filing shows nearly $17 billion in Q3 revenue, up 1,127% year over year, with operating cash flow positive.
Adobe released a local MCP server that lets ChatGPT, Claude, and Gemini work inside Photoshop, Illustrator, After Effects, and Premiere. If you already pay for Creative Cloud, repetitive design work may soon be something you can hand off.
Forbes laid out a six-week holiday plan for the best-return AI uses: product copy, product images, and inventory planning, plus where AI still falls flat. Black Friday is closer than it feels.
New AI agents are shipping for construction, accounting, property management, healthcare, and marketing, built around things like job costing and billing codes. If you're in one of those fields, it's worth ten minutes to look.
-Scott
SmartOwner is published (almost) daily by the team at DigitalTreehouse. Want AI consulting or automations for your business? Reply to this email.
