In partnership with

What's Actually Happening

Anthropic just moved into the office. Today it launched Claude Sheets and Claude Whiteboards in beta for every paid plan, made Claude Data generally available, and showed off Claude Meetings, an agent that joins your Google Meet calls and works while you talk.

It lands days before Anthropic's expected stock market debut, and it's aimed squarely at the tools Microsoft and Google sell to nearly every office.

Also tonight: Google opens Gemini 4 Carbon to every paying developer (at a price that doubles in three weeks), Meituan drops the weights for a 1.6-trillion-parameter open model, and four tools worth trying.

Claude Sheets and Claude Whiteboards are live in beta today on all paid Claude plans, on web, desktop and mobile. Sheets builds and edits real spreadsheets with tabs, formulas, formatting and dropdowns, often from a single prompt, and outlines every change it makes so your team can follow along. Whiteboards turns a messy plan into a shared board with a timeline, owners and dates, groups everyone's notes by theme, and can convert the finished board into a doc or deck.

Claude Dashboards also graduates to general availability under a new name, Claude Data. It connects to warehouses like BigQuery, Redshift and ClickHouse and turns your numbers into dashboards, written data stories or narrated videos. The headline-grabber is Claude Meetings: Claude joins a Google Meet call as a visible participant, drafts the plan while people are still talking, and leaves notes and action items in Claude afterward. TechCrunch reports it's in beta for Team and Enterprise customers, while Anthropic's own overview still calls it "coming soon," so expect a staged rollout.

Our take: Sheets is the one to try first. Anthropic itself says it's simple for now and points power users to Excel export or Claude for Excel, so don't plan to migrate your finance model. Where it shines is the zero-to-first-draft step: a hiring budget, a campaign tracker or a launch plan built from one prompt and then edited together. The bigger idea is shared context, since an idea can move from a whiteboard to a spreadsheet without re-explaining it. On Team plans all three are on by default; Enterprise admins have to switch Sheets and Whiteboards on.

Will you try Claude Sheets or Whiteboards this week?

Login or Subscribe to participate

🏢 Why It Matters: Anthropic Comes For The Office Suite

Spreadsheets, whiteboards and meeting notes are the daily bread of Microsoft 365 and Google Workspace. Anthropic now sells all three inside Claude, backed by a marketplace of more than 9,000 connectors and plugins, and partner apps like Adobe, Atlassian, Box and Notion can now open right inside the chat. OpenAI made its own push into workplace tools a few weeks ago, TechCrunch notes, so the office is now a fight between the AI labs and the incumbents they run on.

The timing isn't subtle. Anthropic is aiming to go public as soon as the week of November 16, with AFP reporting a raise of up to $100 billion at a valuation above $1 trillion, down from earlier talk of $2 trillion (via Malay Mail). Its filings show $33 billion in revenue from January to September, per Gizmodo. Products that turn Claude from a chatbot into the place a team's work actually lives are exactly the recurring-revenue story investors want to hear. Reported IPO figures have shifted all month, so treat them as estimates until pricing.

A Word From Today's Sponsor

The AI Work Handbook That Cuts Your Workday in Half

The 8-hour workday is becoming a 4-hour workday for people who know how to use AI.

Everyone else is still catching up.

This AI work playbook shows you exactly how to cut your work hours in half using AI.

Sign up for Superhuman AI and get:

  • 50+ step-by-step AI tutorials to cut your workload in half — covering every part of your workday, from emails to strategy, used by 1M+ professionals at Google, Microsoft, and NASA

  • Superhuman AI newsletter (4 min daily) so you keep discovering new AI tools and skills to stay ahead in your career — the playbook is just the start

Let’s Talk MODELS! 🤖

Two weeks after handing Gemini 4 Carbon to cybersecurity defenders first, Google opened it today to all paid Gemini API developers in AI Studio, to businesses through Gemini Enterprise, and to Google AI Ultra subscribers in the Gemini app, Antigravity and AI Mode in Search. Google says it ranks first on CursorBench 4.0, Perplexity WANDR and Zapier AutomationBench at more than 2x lower cost per task than the next best outside model.

The catch is the price tag. Carbon costs $2 per million input tokens and $10 per million output through November 30, the same as Gemini 4 Argon, then doubles to $4 and $20. Google also reports a 92% pass rate on its agentic honesty test, up from 70% for Argon, and is previewing a new /swarm multi-agent mode in Antigravity.

Our take: the next three weeks are the cheapest Carbon will ever be. If you build agents, run your evals now at intro pricing so you know whether it earns the doubled rate in December. Every score here is Google's own.

The Open-Weights Giant

Meituan published the open weights for LongCat-2.5 on November 9 under the MIT license, a week after introducing the model. It's a mixture-of-experts model with 1.6 trillion total parameters (the Hugging Face metadata lists 1.8T) and about 48 billion active per token. It's built for long-horizon agent work and reads images and video as well as text.

On Meituan's own numbers it scores 88.4 on Terminal Bench v2.1, up from 70.8 for its predecessor, and 64.6 on DeepSWE v1.1, up from 9.7. It still trails the top closed models on the newer Terminal Bench v4.0, 38.5 against Claude Opus 5's 53.9. The hosted API is on limited-time promo pricing of $0.30 per million input tokens and $1.20 per million output, per VentureBeat.

Our take: MIT-licensed weights at this scale are a gift to anyone who must self-host, but 1.6T parameters means serious hardware. For most teams the cheap API is the realistic way to kick the tires. Run your own agent evals, since every score above is self-reported.

Top 5 In AI Research 🔬

The other stories worth your attention:

DeepMind's AI formally verified Signal's core encryption code, down to the compiled code running on devices, in weeks instead of the years expert teams usually need. See how many lines of proof the agents wrote. (Google DeepMind)

Businesses can now claim verified profiles inside ChatGPT, and some US firms can launch their own customer-facing agents. Here's how it ties into ChatGPT Ads. (TechCrunch)

New York's AI safety law now has a regulator, and OpenAI and Anthropic have to register by January 1. See what each violation can cost. (TechCrunch)

Notion says some Notion AI chats were exposed in a breach at its AI vendor Fireworks. Check whether your workspace was hit and what to rotate. (Notion)

Apple is reportedly testing an app that turns Macs into a private AI server, with a trick for running models too big for one machine. Unconfirmed: it was found by a leaker and Apple hasn't announced it. (9to5Mac)

🛠️ Tools That Are Hot Right Now!

  • 🧭 Modern Web Guidance - free, open-source skills from the Chrome team that stop your coding agent (Claude Code, Copilot CLI, Antigravity and more) from writing legacy web code. One command installs it.

  • 🎮 Tryll Engine - run language, speech-recognition and voice models on the player's own GPU in Unity or Unreal games, no cloud bill. Free open beta; see when the royalty kicks in.

  • 🔎 Reddit Reviews - AI summaries of what Redditors actually think about a product, with likes, dislikes and links to the source threads. Rolling out to US users on iOS, Android and web.

  • 🧩 Create My Widget - describe a home-screen widget in plain words, like a trip countdown or spending tracker, and your Pixel builds it. Free in the November Pixel Drop on Pixel 10 and 11 phones in select markets.

What's The Recap?

Anthropic launched Claude Sheets and Claude Whiteboards in beta on every paid plan, made Claude Data generally available and showed off Claude Meetings for Google Meet, a direct shot at Microsoft's and Google's office suites days before its expected IPO. Google opened Gemini 4 Carbon to all paying developers at intro pricing that doubles after November 30, and Meituan released MIT-licensed weights for its 1.6-trillion-parameter LongCat-2.5. The thread: the AI labs no longer want to be a tab you visit. They want to be where the work happens.

Login or Subscribe to participate

Stay building. 🤖

Our newest video is up. Watch it below.

📬 Read by 14K AI readers. Know someone who'd like it? Send them here to subscribe.

Recommended for you

View all
caret-right