Two companies you may not think about every day just reshaped how AI works at the office

Two companies you may not think about every day just reshaped how AI works at the office

Quick Reads

• Anthropic has released its Agent Skills technology as an open standard, meaning any AI platform — not just its own Claude can now adopt it.
• Enterprise customers on Claude’s Team and Enterprise plans can now manage AI skills centrally, controlling which automated workflows are available across their organisations.
• Partner-built skills are already available from Atlassian, Figma, Canva, Stripe, Notion, and Zapier at launch.
• Atlassian separately announced new AI tools inside its Confluence software, including a visual creation tool and connections to third-party AI apps.
• The two announcements together reflect a wider industry push to embed AI directly into the software workers already use, rather than asking them to switch to separate AI tools.

Anthropic, the company behind the Claude AI assistant, and workplace software giant Atlassian both made significant moves this week to push AI deeper into how businesses actually operate not through flashy demos, but through the tools people open every single morning.

Anthropic announced it would release its Agent Skills technology as an open standard, meaning any company or developer can now build on the same framework, not just Claude users. The San Francisco company also launched organisation-wide management tools for enterprise customers and a directory of partner-built skills from companies including Atlassian, Figma, Canva, Stripe, Notion, and Zapier.

Agent Skills, in plain terms, are sets of instructions that teach an AI assistant how to carry out a specific job, not just answer a question, but complete a repeatable workflow. Think of it like training a new employee, not just on what your company does, but exactly how your team handles expense reports, project updates, or customer queries. A general-purpose AI like Claude might understand what a pull request is, but without a skill, it has no idea how your specific team reviews code or handles quarterly reporting.

Anthropic’s product manager, Mahesh Murag, confirmed that the specification and a reference toolkit are publicly available at agentskills.io and that Microsoft has already integrated it directly into VS Code, its popular coding software. OpenAI, Cursor, and GitHub are also among the early adopters.

The move mirrors Anthropic’s earlier release of the Model Context Protocol, an open standard for AI tools to communicate with outside software, which has since become the de facto communication standard for the AI agent industry, with control handed to the Linux Foundation earlier this year. With Agent Skills, Anthropic is betting the same playbook works twice.

On the same day, Atlassian announced new AI features inside Confluence, its widely used document and knowledge-sharing platform. The company launched Remix, a visual tool now in open beta that turns information stored in Confluence into charts, graphics, and other visual formats without the user having to open a separate application. Atlassian also announced three new third-party agents: one connecting to Lovable to convert product ideas into working prototypes, one linking to Replit to turn technical documents into starter apps, and one working with Gamma to build presentation slides.

For the average worker, the practical effect of both announcements is that the software they use for documents, design, payments, and project management is now able to do significantly more on their behalf automatically, and within the same window they already have open. Enterprises concerned about AI governance can now also manage centrally which capabilities their AI systems have, who controls them, and what guardrails are in place.

The bigger picture is a race among major technology companies to make AI invisible infrastructure rather than a standalone product. Atlassian’s push follows a similar move it made in February, when it added AI agents to its project management software Jira, allowing AI and humans to work side by side on tasks. Salesforce, OpenAI, and Microsoft are all running the same race through their own existing platforms.

The question that remains unanswered is adoption. Open standards only become standards when enough of the industry uses them and Anthropic is competing with well-resourced rivals who have built their own versions of the same idea.