The AI assistants moved further into two places growth companies care about this week: where buyers discover vendors and where employees do their work. Here is what matters today:
The common thread: decide this quarter how your company shows up inside the assistants, because customers and staff are already starting there.
Developments
Go-to-Market
OpenAI Opens a Business Listing Inside ChatGPT, With Mention Counts, Product Feeds and a Branded Agent (November 10, 2026)
OpenAI launched Businesses in ChatGPT in beta on November 10. Companies can claim a verified profile, upload product feeds with prices and availability, and see a Business Hub dashboard of mentions, profile views and website clicks. Eligible U.S. e-commerce and online-service firms can add a labeled Business Agent that answers shopper questions from material they supply, and brands can now pay to make that agent the landing spot for ChatGPT ads. OpenAI says it does not share user conversations, so the data shows how often you were named but not which prompts did it.
So what
This is the first official read on how ChatGPT presents your company, and claiming a profile currently costs nothing. Assign an owner this week to verify the profile, correct the facts and load current pricing, because an assistant that guesses your price is now a sales risk. Treat the dashboard as directional: counts are aggregated and appear only above an activity threshold. B2B firms outside “online services” should expect agents later, so the profile is the near-term win.
Confidence: Established fact — launch reported consistently across seven newsletters; OpenAI’s page could not be fetched, so details were checked against The State of Brand’s write-up.
Go deeper OpenAI · The Rundown AI
Competition
Anthropic Adds Sheets, Whiteboards and Data to Claude and Lets Partner Apps Run Inside It (November 10, 2026)
On November 10, Anthropic released Claude Sheets and Whiteboards, renamed its Dashboards beta as Claude Data, and expanded interactive connectors so apps such as Adobe and Atlassian open fully inside Claude. Plugins from Salesforce (beta), NetSuite and Workday (beta) ship with prebuilt skills for sales, finance and HR work. Claude Meetings, which joins Google Meet calls as a visible participant and drafts plans and action items, is announced as coming soon for Team and Enterprise. The Deep View reports interactive connectors are open to all plans, including free.
So what
The assistant is becoming the place where work starts, and your product may increasingly be opened from inside it rather than from your own login page. If you sell software, find out whether you can ship an interactive connector or plugin; Anthropic’s marketplace already lists more than 9,000. If you buy software, check which spreadsheet, whiteboard and meeting-notes seats overlap with Claude before 2027 renewals. Pilot Sheets on one recurring report before changing anything.
Confidence: Established fact — from Anthropic’s November 10 announcement; plan eligibility for some tools is as reported by The Deep View and The Neuron.
Go deeper Anthropic · The Deep View
Product & TechnologyDeveloping
Gemini 4 Carbon Opens to All Paying Developers at $2/$10 per Million Tokens, but the Price Doubles After November 30 (November 10, 2026)
Two weeks after a restricted release to cyber defenders, Google made Gemini 4 Carbon available to all paying API customers and Gemini for business on November 10, with access rolling out to AI Ultra subscribers. The introductory price of $2 per million input tokens and $10 per million output tokens matches Gemini 4 Argon, then rises to $4 and $20 after November 30. Google says Carbon ranks first on CursorBench 4.0, Perplexity’s WANDR and Zapier’s AutomationBench at more than 2x lower cost per task than the next best external model. It also previewed /swarm, which splits one prompt across a team of agents.
So what
Several newsletters reported only the launch price, so budgets built on it will be off by half from December. If your team tests Carbon this month, model costs at the $4/$20 rate and compare cost per completed task, not list price. Watch /swarm closely: each worker is a separate agent, and Google notes swarms use quotas much faster than single conversations.
Hype check
The benchmark rankings are Google’s own or run with partners. Every’s seven testers titled their review “Gemini 4 Carbon’s Coding Outpaces Its Judgment.”
Confidence: Established fact — pricing and availability come from Google’s post; performance claims are vendor-reported.
Go deeper Google · Every Vibe Check
BusinessDeveloping
Anthropic Starts Its IPO Roadshow: 825 Million Shares at $70 to $78, Listing on Nasdaq as ANT (November 10, 2026)
Anthropic launched its roadshow on November 10 for 825 million Class A shares at $70 to $78 each, roughly $58 billion to $64 billion at the range, with Goldman Sachs and Morgan Stanley named on the prospectus. About 801 million of the shares are new, so most proceeds go to the company. The Neuron reports the amended filing also lists a $26 billion private placement from Nvidia, Google, Microsoft and four other suppliers. The Rundown says the filing states Anthropic now considers its AI R&D capability threshold met, based on what it expects from upcoming models.
So what
A public Anthropic will report quarterly, giving Claude customers a regular view of its margins, compute commitments and pricing pressure. That is useful leverage when you negotiate multiyear AI contracts in 2027. Investors should note that the labs’ largest suppliers are now also among their largest backers. Whatever happens at pricing, keep at least one alternative model qualified for your critical workflows.
Confidence: Established fact — share count, range and ticker are from Anthropic’s notice; the private placement and threshold language come from newsletter readings of the SEC filing and were not verified directly.
Go deeper Anthropic · The Neuron
Perspectives
Go-to-Market
Euclid: Before Hiring More Reps, Check Whether Revenue per Human Hour Holds Up (November 10, 2026)
Euclid’s recap of its Verticalist Summit sales panel argues vertical AI companies should grow ARR per unit of sales effort before adding headcount. Its example is Lassie, where, per Kyle Norton, $15,000 to $20,000 contracts close in six to seven days with a couple of calls and no solutions engineer, and a rep can carry $5 million to $6 million of ARR. The test: take your last ten closed deals and divide each contract’s annual value by the human hours spent winning it, including sales engineering and legal time. If that number falls as you add reps, fix pricing, packaging and tooling first.
So what
This is a practical gate for 2027 hiring plans, at a time when AI tools are changing how much one rep can carry. Run the calculation on Q4 closed-won deals before approving new quota-carrying headcount. If revenue per hour is flat or falling, the cause is usually deal complexity or packaging rather than capacity. The panel also recommends using customers as live sales references, which Euclid says can be worth several times their contract value early on.
Confidence: Informed perspective — a venture firm’s panel recap with company-reported figures; useful as a heuristic, not a benchmark.
Go deeper Euclid Ventures · TLDR Founders
Resources
Learning
Turn Your Corrections Into a Personal Benchmark Before You Switch Models (November 11, 2026)
Every’s head of evals, Mike Taylor, showed at its Thesis conference how to turn the corrections you already make into yes-or-no checks that an AI judge can apply to future outputs of the same task. When he ran one deck task across models, GPT 6.1 Sol reused a brand-guide image and failed his “draw a custom diagram” check, while Grok 4.6 passed. He warns that checks can overfit, turning a fix for one output into a rule that breaks reasonable variations. Every published a copy-paste prompt that converts spoken feedback into a saved checklist.
So what
With Fable 5.2 and Gemini 4 Carbon both new this week, vendor benchmarks will not tell you which model handles your team’s recurring work best. Pick two or three tasks your leadership repeats weekly, such as a board update or account brief, and build a checklist from your own edits. Re-run it whenever you consider switching models or plans, and review the checks monthly for overfitting.
Confidence: Informed perspective — a practitioner method supported by anecdotal model comparisons.
Go deeper Every: How to create your own personal AI benchmark · Every
Go-to-Market
AI Chats Are Adding Vendors to Software Shortlists, and Buyers Ask Them About Price: Two Checks for Your Site (November 9–10, 2026)
A Wynter survey of 200 software buyers, reported by Kevin Indig, found that 62.7% of those who already had vendors in mind added another one through an AI chat. In those chats, 57.5% asked about pricing and 54.5% compared vendors; afterward, 72.5% visited vendor websites and 48% searched Google. Separately, Resend serves its pricing page as a markdown table when an agent adds “.md” to the URL, so a coding agent can read the price for a given volume in one request, according to Latent Space.
So what
Part of your shortlist position is now decided in conversations you cannot see. Ask two or three assistants what you cost at a typical customer’s volume, then fix what they get wrong on your site and in your new ChatGPT business profile. If your pricing says “contact us,” expect AI answers to fill the gap with guesses. A machine-readable pricing page is a cheap fix for product-led and usage-priced businesses.
Confidence: Informed perspective — a small, unweighted survey run by a firm Indig has invested in, plus one company’s practice.
Go deeper Growth Memo · Latent Space
