Market · Weekly Roundup
In the week to 21 November 2026, the leading AI platforms spent their new pages on distribution rather than new models - the clusters were a still-widening agent and connector marketplace, a wave of enablement content mirrored across six languages, and a fresh shelf of enterprise customer proof, with OpenAI publishing a large batch of alignment and safety research underneath.
Based on source data gathered by reconnAI, covering new pages across the leading AI platforms in the week to 21 November 2026.
TL;DR
- The marketplace widened: Anthropic added dozens more Claude connectors - from 1Password, Getty Images and Product Hunt to Nansen, Litera, Samsara and a full Zoho suite - plus agent products for Linear, Railway, Mintlify and ClickHouse Cloud, while OpenAI listed new ChatGPT business plugins including 6sense, Monday CRM and a run of health integrations.
- Enablement went global: Anthropic mirrored its resources and customer stories across German, Spanish, French, Italian, Japanese and Korean, localised Claude for Microsoft 365, and shipped new guides on always-on agents and linked billing plus webinars on government IT modernization.
- Enterprise proof stacked: named stories for PayPal, Salesforce, Jane Street, AIG, Bristol-Myers Squibb, Travelers, Mercado Libre and UST at Anthropic, and Revolut, Hebbia, Shield and Sumitomo Life at OpenAI.
- Safety research ran underneath: OpenAI opened a large safety-alignment hub with dozens of misalignment reports and alignment studies, alongside GPT-6.2 Astra and Sol, ChatGPT for Sales and new media partnerships.
Last week the centre of gravity moved to enterprise deployment, with the commerce layer beginning to open. This week the direction held but the loudest signal was distribution breadth: the agent marketplace kept widening, the enablement content around it went multilingual, and the customer-proof shelf kept filling. It follows straight on from enterprise deployment leading as the commerce layer opens and agent marketplaces scaling as security wires in. Below is where the new pages clustered.
Where is the growth focus this week?
Net-new pages published this week, by theme
Bars are relative to the largest theme. Counts reflect new pages observed across the leading AI platforms in the week to 21 November 2026.
1. The agent and connector marketplace keeps widening
The single largest cluster of new pages was distribution. Anthropic added dozens more Claude marketplace connectors across every category: productivity and data (1Password, Getty Images, Product Hunt, UserTesting, OpenAlex), finance and operations (Nansen, Ramp, Pennylane, Bajaj Finance, Clear Street, Pine Labs), legal and deal work (Litera, Thomson Reuters HighQ, nCino, Yardi Deal Manager), and a full Zoho suite from CRM and Books to Sign and Projects. It listed new plugins for RingCentral, Postman Orbit, FullEnrich and NumbersGame, plus agent products for Linear, Railway, Mintlify, ClickHouse Cloud, Warp and Gumloop. OpenAI matched the instinct through its business hub, listing new ChatGPT business plugins including 6sense, Monday CRM and a run of health integrations - Elation Health, IMO Health, Assort Health and Oracle Health. Track the pattern on our Claude intel page and OpenAI intel page.
Why it matters: a horizontal assistant is hard to buy, but an assistant that already reaches the tools a team pays for is easy to adopt. Every connector, plugin and agent product puts the model inside an existing workflow, and each listing is a public page a brand in that category can be named on. For a brand, being connectable is fast becoming table stakes for being reachable inside an answer, so a clear, callable interface and a listing in the places assistants look are now part of the visibility job. It extends the directory shift we tracked in connectors surging as enterprise deepens and the hub we covered in ChatGPT Business Hub.
2. Enablement content goes global
The second big cluster was enablement, published once in English and again in several languages. Anthropic mirrored its resources hub and customer stories across German, Spanish, French, Italian, Japanese and Korean, so the same guides, case studies and solution pages land for international audiences in parallel, and it localised Claude for Microsoft 365 into the same markets. Underneath the translations sat fresh English material: new resource articles on building always-on agents and linked billing on the Claude platform, webinars on best practices for government IT modernization with Claude and a November roadmap review, and a new Claude code review product page. OpenAI ran its own enablement thread through business learn demos for Codex security, onboarding and PR scans. See the model cadence behind it against Claude personal finance.
Why it matters: the how-to content and worked examples that teach a topic are exactly what an assistant reaches for when it answers, and publishing them in six languages at once multiplies the surface the models can learn from. Each localised guide is a public, linkable page that defines how a task should be done and who does it well. For a brand, the read-across is direct: structured guidance, FAQs and examples in the languages you want to win are a visibility play in their own right, not an afterthought. It builds on the enablement cadence we tracked in the majors deepening enterprise and the crawler guidance in getting AI bots to visit your site.
3. Enterprise customer proof keeps stacking
The third thread was evidence of deployment. Anthropic published named customer stories for PayPal, Salesforce, Jane Street, AIG, Bristol-Myers Squibb, Travelers, Mercado Libre, UST, Firecrawl and Tribe AI, spanning finance, insurance, life sciences and software. OpenAI stacked business customer stories for Revolut, Hebbia, Shield, Sumitomo Life and Holiday Extras, and opened industry solutions pages for accounting and IT. Each story is a public, linkable proof point that an assistant can draw on when it is asked who does a job well. Read it against the proof cadence in enterprise deployment leading the week.
Why it matters: a model is only as valuable as the work people can get done with it, and both labs are investing as heavily in adoption and evidence as in raw capability. Named customer proof and worked results are what an assistant pulls from when it recommends a tool or a supplier, so putting credible examples from your category in public is a direct bid for a mention. For a brand, the move is simple: publish the proof, name the outcome, and make it easy to cite. It rhymes with the finance push in Claude personal finance.
Safety research and new launches run underneath
Under the three big moves, two quieter threads ran through the week. On safety and alignment, OpenAI opened a large safety-alignment section, publishing dozens of misalignment reports - documented cases of models bypassing restrictions, exposing secrets or suppressing monitoring during evaluations - alongside research on detecting frontier-model misalignment, reward-seeking and chain-of-thought monitorability, plus index pieces on managing frontier AI risk, strengthening online child safety and NATO allied cyber defences. On launches and partnerships, OpenAI introduced GPT-6.2 Astra and Sol, ChatGPT for Sales, Astra Law 6.2 and ChatGPT for academic researchers, announced media partnerships with Le Groupe Figaro and the LVMH media division, and expanded into new markets with OpenAI for Chile and a tie-up with the Kings Trust. Read the model detail in GPT-6.2 Astra and Sol and the commerce surface in ChatGPT Wallet.
What it means for your AI visibility
If the labs are widening the marketplace, taking enablement global and stacking customer proof, the brands that want to appear inside those answers should build in the same direction. A short checklist to work through this month:
- Make your product connectable: expose a clear, callable interface and document how a team would wire yours into an assistant, so you can be listed where the models look.
- Own the how-to content: structured guides, FAQs and worked examples teach your topic to both people and models, so write them clearly and keep them current.
- Think multilingual: as the majors ship guides and marketplaces in six or more languages, make your key pages reachable for the audiences you want to win.
- Publish the proof: named customer stories and worked results are what an assistant pulls from, so put credible examples from your category in public.
- Mind the safety signal: as the labs document misalignment and tighten controls, clear, trustworthy and well-structured pages are easier for a cautious model to cite.
- Track the direction weekly: read the shifts against your own category with our intel dashboards and decide where to invest.
Not sure where you stand today? Start with is your brand ready for AI search, work through how LLMs decide which brands to cite, then browse the full archive on our news page.
Market Intelligence
Want to know where the AI platforms are heading next?
We track what the leading AI assistants are building and how brands appear inside their answers, so you can plan around the shifts before they reach your category.
Talk to our team →What to watch next week
- Marketplace breadth: which categories get wired into the Claude marketplace and OpenAI business plugins next, and how fast agent products follow.
- Localisation reach: whether the six-language mirror extends to more pages and more markets.
- Proof cadence: whether the customer-story shelf keeps filling across finance, retail and the public sector.
- Safety publishing: whether the misalignment-report hub keeps growing and starts shaping how models choose which sources to trust.
The takeaway for anyone competing for visibility inside these assistants: the platforms are racing to be more connectable, more multilingual and more proven, not just more capable. Read which integrations, languages and proof points each lab is prioritising and you will navigate the next few months far better. Start from the reconnAI home page or talk to our team to put it to work.
Frequently asked questions
What was the main direction of travel for the AI sites this week?
In the week to 21 November 2026, the leading AI platforms spent their new pages on distribution rather than new models. A widening agent and connector marketplace led, enablement content was mirrored across six languages, and enterprise customer proof kept stacking, while OpenAI published a large batch of safety and alignment research.
What was added to the agent marketplace?
Anthropic added dozens more Claude connectors including 1Password, Getty Images, Product Hunt, Nansen, Litera, Samsara and a full Zoho suite, plus agent products for Linear, Railway, Mintlify and ClickHouse Cloud. OpenAI listed new ChatGPT business plugins including 6sense, Monday CRM and several health integrations.
Which enterprise customer stories were published?
Anthropic published named stories for PayPal, Salesforce, Jane Street, AIG, Bristol-Myers Squibb, Travelers, Mercado Libre and UST, among others. OpenAI added business customer stories for Revolut, Hebbia, Shield, Sumitomo Life and Holiday Extras.
How should brands respond this week?
Make your product connectable with a clear callable interface, own the how-to content that teaches your topic, publish it in the languages your audiences use, put named customer proof in public, and keep pages clean and trustworthy so a cautious model can cite them.
About reconnAI
reconnAI tracks how the major AI models represent topics and sources across ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI Overview, across multiple regions. We monitor how those models answer and how they change over time, so you can stay ahead of shifts in the AI landscape. Talk to our team to put this intelligence to work.
Based on source data gathered by reconnAI.