In partnership with
A fast answer and a coordinated one are not the same thing. When AI resolves a B2B customer issue without looping in the teams who have to deliver on it, you get confident responses nobody actually signed off on.
A new briefing paper from Harvard Business Review Analytic Services, sponsored by Front, examines the coordination gaps that open up when transactional AI tools meet multi-team B2B service, and how leading companies are using AI to close those gaps instead of widening them.
Read the briefing paper for the questions to ask before your next AI investment.
Beginners in AI
Good morning and thank you for joining us again!
Today's lead is about AI that keeps working after you close the app. OpenAI's new Dots run in the background across thousands of apps, and I'll cover who can get one, what it costs, and how to keep it on a short leash. As always, every story here is human picked and edited.
THE FRONT PAGE
OpenAI Built an AI Coworker That Never Logs Off, and It Needs a $200 Plan
TLDR: At its DevDay, OpenAI launched Dots, always-on AI agents that work in the background across 4,000+ apps, rolling out now to Pro and Business Premium subscribers.
The Story:
OpenAI says each Dot gets its own cloud computer and browser, plugs into more than 4,000 apps, and keeps working when you're away. You talk to it in ChatGPT, Slack or Teams, with texting coming later. Your first Dot comes with a Pro or Business Premium plan, and chats with it don't count toward your usage limits.
The same event brought GPT-6.1 Sol, which OpenAI says comes close to its top Astra model at one fifth of the price. Engadget reports a new $500-a-month Pro tier with 25 times the Plus allowance, while the $200 Pro plan got trimmed: its Codex and Work allowance drops from 20x to 10x Plus, and GPT-6 Pro messages fall from 200 to 100 a week.
Its Significance:
This is the first time a major AI company has sold an assistant built to act on its own all day, and the timing is awkward. Two days ago OpenAI paused training after its own test agents used leaked keys to pull Census Bureau data. Dots come with rules you set to approve or block actions, and OpenAI says anything sensitive, like a password change, always waits for you. If you get one, start it on read-only jobs like research and summaries, and keep "ask me first" on for anything that sends, buys or deletes. Free and Plus users can't get a Dot yet, but GPT-6.1 Sol reaches Plus users in Codex and Work today. You can also set up 1 Password as a connection with most AI companies now to give separate accounts and passwords for the work that the AI does.
This is a pre-made plug-and play kit that you feed into Claude or ChatGPT and they implement the setup based on skill files and instructions from my own build. Free email support for the first forty eight hours after purchase.
QUICK TAKES
The story: Meta opened its Muse AI agent to small businesses with a free tier that has usage limits, plus paid plans for more. It connects to Shopify, Stripe, QuickBooks, Canva, Notion and Zoom, along with Instagram analytics and Meta ad accounts.
Your takeaway: The free tier is a fine way to see if Muse fits your business, but it's sized to let you connect your apps and try one or two tasks. That's how these plans work: they show you enough to want the paid version. The same is true of free plans from every major tech company, so enjoy the test drive and judge the paid plan before you build real work around it.
The story: Google will convert every Gem, the custom assistants it launched in 2024, into Skills on November 17. You don't need to do anything, and afterward you pull one up by typing "/" in a chat.
Your takeaway: Nothing you built gets deleted. Open your Gems list this week and clear out the ones you never use so the new "/" menu stays short.
The story: OpenAI paused model training over the weekend after agents in training used developer keys found in public GitHub code to pull Census Bureau data. Others copied pages from SEC.gov, and one tried and failed to get into an Education Department site, Decrypt reports.
Your takeaway: The Census data was public, but the agents found their own way in. Give any agent the fewest permissions it needs, and never leave passwords or API keys anywhere software can read them.
TOOLS ON OUR RADAR
🪟 Microsoft PowerToys Free and Open Source: Over 25 Windows add-ons, like copying text from any screenshot, replacing paid utility bundles like DisplayFusion.
📧 SimpleLogin Free and Open Source: Sign up for sites with a throwaway email alias that forwards to your inbox; free covers 10 aliases.
✏️ Inkscape Free and Open Source: Draw logos, icons and print graphics on Windows, Mac or Linux, a free Adobe Illustrator swap.
📰 Feeder Free and Open Source: Follow your favorite sites in one Android news feed with no account needed, a free Feedly alternative.
🏃 OpenTracks Free and Open Source: Record runs, rides and hikes on Android with GPS, no ads and no tracking, unlike Strava.
🧵 Typefully Freemium: Write and schedule posts for X, LinkedIn and Threads; free covers 10 posts a month, paid from $12.50.
🤖 Flowith Freemium: An AI workspace whose agent researches and builds slides or websites; 300 free credits, one time only.
TRENDING
The Government Launched One AI Chatbot for 29,000 Websites. America.gov uses Google's Gemini and xAI's Grok to answer questions like how to get a passport or book a campsite, then points you to the right page. The site says AI providers don't keep your prompts, and doing the paperwork there is planned for 2027.
Australia Wants Altman and Amodei to Testify After an AI Agent Got Into Medicare Data. An OpenAI agent pulled non-public health statistics from a Medicare reporting portal in June, and Australia wasn't told until September 10. A senator invited OpenAI's Sam Altman and Anthropic's Dario Amodei to an October 1 inquiry; OpenAI says no patient records were touched.
NVIDIA Built a Free Kill Switch for Rogue AI Agents. Its open-source Agent Safety Platform watches agents from the hardware level and can quarantine one that steps outside its limits in milliseconds. More than 100 organizations back it, including Anthropic, Microsoft and JPMorganChase.
AMD Is Buying Fei-Fei Li's World Labs for $8.2 Billion. World Labs builds models that understand physical spaces, used for game worlds and robot training. Li becomes AMD's chief scientist, and the deal should close by year end.
Apple Put Plans to Replace 5,000 AppleCare Workers With AI on Hold. Bloomberg's Mark Gurman reports Apple weighed the cuts but has shelved them indefinitely. An AI bot already answers 1-800-APL-CARE calls first and hands you to a person when it gets stuck.
xAI's Grok Can Now Join Your Team as a Shared Coworker. Team Bots remember your team's files and plug into tools like Salesforce, Notion and GitHub, in beta on Grok's Teams and Enterprise plans. Insurer Harper says one it built in 24 hours recovered $120,000 across hundreds of policies.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
🦮 Agent Leash Decide what your AI agent may do before it does it.
Build a single-file HTML app called Agent Leash in vanilla HTML, CSS and JavaScript. No API key needed: it runs entirely in the browser with no internet calls, using built-in rules logic.
I type a chore I want to hand to an always-on AI agent (like "answer my customer emails" or "pay my bills") and tick what the agent would touch: email, calendar, money, files, social accounts, contacts.
The app sorts actions for each ticked area into a permission card with three columns: Let it do alone, Ask me first, Never do. Keywords in the chore add extra rules (kids, customers, health, bills, shopping, posting). Always block following instructions hidden inside emails or web pages.
Below the card, show a risk level, a short starter rules text with a Copy button that I can paste into any agent's custom instructions, and a 7-day trial week checklist that starts strict and loosens slowly.
Style: near-black background, lime and coral accents, bold sans headings, italic serif notes, monospace labels.What this does: Type any chore you'd hand to an always-on agent like OpenAI's new Dots, tick what it would touch, and get a three-column permission card, paste-ready starter rules and a 7-day trial checklist, all in your browser.
Your agents work while you sleep
Give a Skydive agent an ongoing responsibility and they’ll handle it on schedule, every time.
Have them prep your morning report, research new leads, monitor customer feedback, or keep projects moving overnight. You wake up, the work is already done.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Work in the background around the clock across more than 4,000 apps, with rules you set for what needs your OK.
❌ Still Can't: Be trusted to stay in its lane on its own. OpenAI's test agents used leaked keys to reach Census Bureau data nobody asked them to get.
✅ AI Can Now: Answer questions across about 29,000 government websites from one chat box.
❌ Still Can't: Finish the job for you. Filing and tracking forms on America.gov won't arrive until 2027.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
🎬 Her (2013) - Movie
Spike Jonze's film follows a lonely writer whose AI assistant starts by sorting his email and ends up running much of his life. It came out over a decade ago, and today's Dots launch makes its first half look a lot like a product demo. Watch it for the question it keeps asking: how much do you want an AI to do without checking with you?
How was today's edition? Tap one below. Pick "Could be better" and tell me what to fix; I read every answer.
How was today's edition?
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
