OpenAI gave businesses a formal way to claim profiles inside ChatGPT. Anthropic turned Claude into a workspace with spreadsheets, whiteboards, and a live data tool. And Microsoft’s speech recognition model is worth a closer look on price and accuracy. Three things worth knowing.
OpenAI Opens ChatGPT to Business Profiles and Agents
Image: OpenAI
OpenAI launched Businesses in ChatGPT on November 10. Eligible businesses where ChatGPT is offered can claim and verify their business, then update key details in a management surface called Business Hub. Retailers can also upload product feeds with pricing and availability, and ChatGPT may use this information to answer customer questions.
The second piece is Business Agents. A Business Agent answers customer questions inside ChatGPT with the business’s product and policy information, using guidelines the business controls. OpenAI says Business Agents are currently limited to eligible ecommerce and online service businesses in the United States, with access expanding to more types of businesses over time.
The third piece is Insights. A claimed business can see how often it is mentioned in ChatGPT and explore discovery topics, the conversation themes where it appears.
Claiming a business requires verification. Eligible Shopify merchants can connect their store and create an agent. Businesses without a Shopify store can still claim their business. A blue check appears on the profile once it is claimed and verified.
One line from OpenAI’s own page is worth reading: “ChatGPT continues to make independent recommendations and comparisons. Claiming does not guarantee higher placement or sales.” That is the right policy. It just changes what this is.
Why it matters: Businesses now have a formal surface to manage how they appear in ChatGPT. The practical question is whether the Insights data gives you anything useful. Mentions and discovery topics are proxy signals, not conversion data. The Business Agents have a clearer near-term use case for ecommerce: a branded bot that knows your product catalog and policies is a better experience than generic answers pulled from your website. For teams tracking brand visibility in AI, this is worth watching early. Set up the profile, monitor mentions for a few months, and see what shows up in discovery topics. The signal is early but the barrier to entry is low.
Find out more: OpenAI’s Businesses in ChatGPT
Anthropic Launches Claude Sheets, Whiteboards, and Data
Image: Anthropic
Anthropic launched three new tools inside Claude on November 10. Claude Whiteboards and Claude Sheets are in beta. Claude Data, formerly called Claude Dashboards, is now generally available. All three are available on paid plans across web, desktop, and mobile.
Whiteboards is a shared board where teams brainstorm with Claude. You can talk through a plan, and Claude turns it into a visual board with a timeline, owners, and dates. The team can edit it in real time. When done, Claude can convert the board into a document or deck, or export it to tools like Lucid, Miro, or Vani by Zoho. Sheets lets Claude build spreadsheets with tabs, formulas, formatting, and drop-downs, often from a single prompt. You can share sheets with teammates and work on them together. Exporting to .xlsx or CSV is available for anything requiring full Excel functionality; partner tools include Airtable, Asana, monday.com, and Smartsheet.
Claude Data connects to data warehouses including Amazon Redshift, BigQuery, and ClickHouse. The GA release adds certified queries, in beta: data teams can certify the queries everyone should use, and charts built on those metrics carry a “certified” badge. Live sharing is also in beta, allowing colleagues to see up-to-date numbers on a shared dashboard even if they don’t have access to the connectors behind it.
Also on November 10, interactive connectors were upgraded. Tools like Adobe, Atlassian, Box, and Notion can now open as full applications alongside a Claude conversation. Plugins from Salesforce (in beta) and Workday (in beta), along with NetSuite, come with built-in skills for sales, finance, and HR work.
Claude Meetings, which will let users add Claude to calls for automated notes and document updates, is announced as coming soon.
Why it matters: Anthropic is making a serious push into enterprise workflows, and the connector upgrade is what makes it different from other AI tools. Opening Adobe or Atlassian in full beside a Claude conversation changes the work surface. Sheets and Whiteboards are early compared to dedicated tools: Sheets is not Excel, and Whiteboards is not Miro. Anthropic says as much: sheets have simple spreadsheet functionality for now, and the export paths and partner integrations are there from launch. Claude Data going GA is the more quietly important move. When a data team can certify the queries behind a dashboard and colleagues can view live numbers without direct connector access, you have something that starts to look like a real BI pipeline rather than a novelty. The question every team should ask before adopting: does this replace a tool you already pay for, or does it add another surface to manage?
Find out more: Anthropic’s Introducing Claude Sheets, Whiteboards, and Data
Microsoft MAI-Transcribe-2 Is Worth a Closer Look on Price and Accuracy
Image: Microsoft
Microsoft released MAI-Transcribe-2 in September 2026. It is in public preview in Microsoft Foundry and through Azure Speech, and it can be tried in the MAI Playground and on Open Router. The model handles speaker diarization, word-level timestamps, keyword biasing, code-switching across commonly blended language pairs, and automatic language identification. It supports 60 languages without requiring the user to specify a language in advance. Transcription style is configurable: a “verbatim” setting captures filler words and false starts for compliance and analysis work; a “clean” setting removes fillers for more readable captions, notes, and transcripts.
The price at launch: $0.10 per hour of audio. Microsoft says that is a limited-time offer through December 31, 2026.
Microsoft says MAI-Transcribe-2 beats Gemini 3.5 Transcribe, GPT-Transcribe, Whisper V3-Large, and ElevenLabs ScribeV2. On the FLEURS multilingual benchmark, Microsoft says the model ranks first across 60 languages with an average word error rate of 5.2%. On Artificial Analysis, Microsoft says it leads the accuracy-latency Pareto frontier and is second on the word error rate leaderboard. Citing evals run by Artificial Analysis, Microsoft reports it is 10 times faster than OpenAI’s GPT-Transcribe, 7 times faster than ElevenLabs Scribe v2, and 5 times faster than Gemini 3.5 Transcribe.
Why it matters: $0.10 per hour is a price worth paying attention to. For teams that transcribe meetings, calls, clinical notes, or legal records, diarization plus word-level timestamps plus configurable output style covers most real workflows in a single model. The multilingual support across 60 languages without manual language selection is useful for global teams. The benchmark claims come from Microsoft’s announcement. Worth remembering when you evaluate against your own audio. Build the evaluation now, while the model is in public preview; the year-end price deadline gives you a reason to move.
Find out more: Microsoft AI’s MAI-Transcribe-2 announcement