Plus a free training library that could replace your AI consultant.

Your AI is in your Messages app now. OpenAI quietly launched an Apple Messages plugin for ChatGPT on Mac (Apple silicon only). Connect your inbox and ChatGPT can search your threads, draft replies, and send texts with your name attached. No AI label visible to whoever gets the message. By default it asks your approval before hitting send, but you can turn that gate off completely. That last part is the part you need to sit with for a minute.

If you run a business where client trust lives in personal communication, this matters right now. Not eventually. Not "when AI gets better." Now. Your staff can enable this today on any ChatGPT plan, and the person on the other end of the thread has no idea they are talking to software. That is a client relationship question and potentially a liability question, depending on your industry. Before anyone on your team turns this on for business communications, you need a policy. "Ask me before sending" as the default is the only sane starting point. Read the privacy policy before connecting any client inbox to it. Here is what OpenAI announced.

Free training for a tool you are already paying for. Anthropic just pulled all of its learning materials into one place at academy.claude.com. 355 tutorials covering Claude basics, Cowork, Code, and the API, from beginner to expert level. No subscription required. This is the kind of thing you would normally pay a consultant a few hundred dollars to walk you through in an afternoon. Now it is free and self-paced. If anyone on your team uses Claude for any regular task, this is the fastest way to get more out of the subscription you are already paying for.

The web is filling up with AI writing, and it is changing where your content gets found. A new study reported by TechCrunch puts the number at roughly one in three new web pages showing detectable signs of AI authorship since ChatGPT launched. That is a lot of noise to compete with. At the same time, Promptwatch data shows Reddit's share of ChatGPT citations dropped 86% since mid-July, from about 4% down to 0.5%. If you built any part of your content or visibility strategy around community platforms showing up in AI answers, that foundation just shifted. Nobody from OpenAI has explained why. The practical takeaway: AI search sources are moving fast, and no single platform is a safe bet right now. Original, human-voiced content is the hardest thing to flood out. Full story at TechCrunch.

Tools Worth a Look

Claude Academy, 355 free tutorials on Claude, Cowork, Code, and the API; beginner to expert; no subscription needed. For any owner or team member already using Claude. Free.

Slack Code, coding agents inside shared Slack channels; anyone on the team can steer a software build in real time, with human approval before deploying. For small teams that live in Slack and want to supervise AI-built software without leaving the platform. Available on all Slack plans.

Microsoft Clarity AI Scrape-to-Referral Report, free report inside Microsoft Clarity showing how often AI bots scrape your site versus how much referral traffic they actually send back. For owners who want to know if AI search visibility is translating into real visits. Free.

Also Worth Knowing

Google launched a "Preferred Sources" button that any publisher can embed on their site; when readers click it, that site gets pinned as a trusted source across Search, Discover, Google News, AI Overviews, and AI Mode. Worth knowing if you publish content.

Forbes has a practical SMB framework for owners who feel overwhelmed: start with one workflow, measure it, then expand. Sensible advice that most consultants charge for.

Zapier ZapConnect on September 23 is a free virtual event built around one honest question: when does AI actually belong in a workflow, and when does it just add friction? Free to register.

Forbes on automating small business marketing with AI, 10 specific approaches and the tools behind them.

OpenAI previewed Private Safety Processing for enterprise, monitors misuse patterns without storing customer data, a direct shot at Anthropic's enterprise privacy positioning.

Meta AI now has a dedicated Mac desktop app as part of what the company is framing as a small business push.

Bredin research finds economic uncertainty is reshaping how small businesses invest in AI in 2026, worth a look if you are trying to benchmark your own pace of adoption.

BizTech put together a practical guide for SMBs on building an AI-ready endpoint strategy, useful if you are thinking about device and security policy before rolling out AI tools to staff.

-Scott

SmartOwner is published (almost) daily by the team at DigitalTreehouse. Want AI consulting or automations for your business? Reply to this email.