Market · Weekly Roundup
The leading AI platforms are the public surfaces where OpenAI, Anthropic and Google show what their models can do, and in the week to 22 August 2026 the new pages pushed agents deeper into partner ecosystems and the developer platform, with a widening run of named partners, computer-use, Skills and Files APIs, and a firming layer of governance, data controls and policy leading the direction of travel.
Based on source data gathered by reconnAI, covering new pages across the leading AI platforms in the week to 22 August 2026.
TL;DR
- Partners lead: OpenAI added Quantium, TCS, Asana, Replit and Stampli pages plus an Nvidia ChatGPT Work page, while Anthropic named League, ABC Legal, monday.com, Slack and JetBrains, turning capability into channel.
- The agent grows up: managed agents, an agent-first product build-out and a computer-use, Skills and Files API release moved the agent from demo to something teams build on.
- The platform widens: Claude Code guides, an AI CI/CD on-call page and a research-workflow Skill pushed the developer surface, shipped across five languages.
- Rules firm up: OpenAI published zero data retention for frontier models, cyber-capability pacing, national-security oversight and a new AI Futures programme, putting governance on named pages.
Last week the leading AI platforms moved from proof to deployment. This week they built on it and put the agent to work through other people. The new pages leaned into partners and the platform: named integrations, managed agents that any employee can build, and a developer surface that now spans computer use, Skills and Files. Underneath, a run of governance and policy pages firmed up the trust layer. It carries straight on from agents go to work across the enterprise and the channel build we tracked in the labs turn trust into enterprise reach. 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 22 August 2026.
1. Enterprise agents reach into partners
The largest cluster of new pages was about reaching customers through partners. OpenAI stood up partner pages for Quantium and TCS, index pages for Asana, Replit and Stampli, an Nvidia ChatGPT Work page and a partnership with codeAI. It added business plugins for Azure DevOps and data, a downloadable ChatGPT Work guide for sales teams and a page on how its own sales team uses ChatGPT Work. You can watch the picture build on our OpenAI intel page.
Anthropic answered with proof and product. A named customer story for League landed alongside a run of build-focused pages: ABC Legal turning every employee into a builder with managed agents, monday.com becoming an agent-first product where humans and agents collaborate, Slack building human-agent teams that turn conversation into knowledge, and JetBrains evaluating and deploying Claude Fable 5. A CrowdStrike Falcon Fusion plugin wired Claude into a security tool teams already run. You can see the shelf grow on our Claude intel page.
Why it matters: partner pages and managed agents are a distribution strategy. Each named partner is a reference a buyer can trust, and each managed agent is a way for a team to build its own workflow without leaving the tools it already uses. It is the productised version of the move that took Claude into Microsoft 365 Copilot and turned ChatGPT into an agent for work.
2. The developer platform widens
The second cluster was about the developer platform. Anthropic released computer use together with the Skills API and Files API in a single page, then shipped it across German, French, Italian, Japanese and Korean in one go. Around it came an AI CI/CD on-call page, a Claude Code guide for startups, a page on maximising the value of your Claude Code sessions, and a use case on packaging a research workflow into a reusable Skill.
The direction is consistent: give developers stable surfaces to build on rather than a single chat window. Computer use lets an agent operate a screen, the Skills API packages a repeatable job, and the Files API gives it something durable to work over. Together they turn the assistant into a platform, extending the model line from Claude Sonnet 5 to Claude Opus 5.
Why it matters: when the same capability ships as an API and as documentation on the same day, it signals that the feature is meant to be built on, not just tried. A team that can call computer use, Skills and Files from its own product will keep the model close to its workflow, and the brands that appear inside those workflows are the ones an assistant can reach. It carries on the wire-up-and-skill-up pattern we tracked in the labs race to wire up, skill up and prove it.
3. Governance and trust firm up
The third cluster was governance. OpenAI published a run of policy and trust pages: zero data retention for frontier models, pacing model development against cyber capabilities, a piece on the defenders window, strengthening democratic oversight in national security, joining the Ports Pike project, new policy ideas for the intelligence age, and a new AI Futures programme with its own news page. Each puts a governance position on a named, browsable page rather than a one-off report.
Anthropic kept pace on the trust side. Its approach to teaching and learning AI set out how it wants the technology adopted responsibly, and the CrowdStrike security plugin tied Claude into the tooling that enterprise security teams already trust. It is the governance conversation we first flagged when safety took the front page.
Why it matters: as models act on real systems, a visible governance and data-handling posture is what carries a platform, and a brand, through procurement. When data-retention controls, cyber-capability policy and oversight commitments become permanent pages, they signal that trust is now part of the product, not a footnote. It extends the security thread we tracked in the plugin shelf fills, security wires in.
Reaching every market at once
Under the three big moves, two quieter themes ran through the week: localisation and enablement. Anthropic shipped its computer-use release across five languages and published certifications pages in German, French, Japanese and Korean, while OpenAI opened ChatGPT to new audiences with a teens experience. Anthropic also built out a Claude Science product page and guide, and a Japanese use case on packaging a research workflow into a Skill. The message is that the labs want their agents understood, adopted and taught in every market at the same time, not only in English.
What it means for your AI visibility
If the labs are wiring agents into partners, widening the developer platform and putting governance on named pages, the brands that want to appear inside those answers should build in the same direction. A short checklist to work through this month:
- Be a named partner: as the labs publish partner and customer pages, being one of the named references is one of the clearest ways to be pulled into an answer.
- Make your product buildable: managed agents and Skills reward products that expose a clear, callable job, so document how a team would wire yours in.
- Publish for the platform: APIs plus how-to pages on the same day show that step-by-step content is easy for a model to lift, so write down how your product solves a task.
- Put governance on a page: a visible data-handling and compliance posture helps a model place you as a safe choice inside an answer.
- Go multilingual: as the labs localise agents and certifications, being present in more than one language widens where you can be cited.
- Track the direction weekly: read the shifts against your own category using 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 building a GEO strategy and 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
- Partner velocity: how fast the named-partner and named-customer runs keep growing, and which sectors get referenced next.
- Managed agents: whether more platforms follow monday.com and ABC Legal in letting any employee build an agent.
- The platform surface: how quickly computer use, Skills and Files move from release to widely built-on, and what OpenAI ships in reply.
- Governance: whether zero data retention, cyber-capability pacing and AI Futures turn into a standing policy shelf that rivals answer.
The takeaway for anyone competing for visibility inside these assistants: the platforms are racing to be partnered, built-on and trusted, not just capable. Read which partners, platform surfaces and governance pages 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 22 August 2026, the leading AI platforms pushed agents deeper into partner ecosystems and the developer platform. The new pages led on enterprise agents and partners (named customers, partners and managed agents), followed by the developer platform (computer use, Skills, Files and Claude Code), with a firming layer of governance, data controls and policy underneath.
Which partners and customers appeared this week?
OpenAI added partner and index pages for Quantium, TCS, Asana, Replit, Stampli, Nvidia and codeAI, plus business plugins for Azure DevOps and data. Anthropic named League, ABC Legal, monday.com, Slack and JetBrains, and added a CrowdStrike Falcon Fusion plugin.
What changed on the developer platform?
Anthropic released computer use with the Skills API and Files API on one page, shipped across five languages, alongside an AI CI/CD on-call page, a Claude Code guide for startups, a sessions guide and a research-workflow Skill. Together they turn the assistant into a platform developers can build on.
How should brands respond this week?
Aim to be a named partner, make your product buildable through clear callable jobs, publish how-to content next to any integration, put a visible governance and data-handling posture on a page, and consider more than one language. Our guide to how LLMs decide which brands to cite walks through the detail.
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.