The news: Anthropic is embedding itself where work actually happens. Two recent launches—Claude CoWork and Salesforce’s updated Slackbot—show a deliberate push into agentic workflows that live inside daily tools, not separate AI apps.

Why it’s worth watching: Unlike Microsoft, Google, Amazon, or Meta, which monetize AI across advertising, subscriptions, licensing, and cloud, Anthropic relies more heavily on subscriptions and licensing, with its estimated 2025 AI revenues of roughly $4 billion. 

That constraint helps explain a two-pronged strategy. On one hand, Anthropic taps Slack’s captive audience, reaching users directly inside team conversations and shared workflows. On the other hand, Claude CoWork extends that agentic capability to the desktop, where it can act on files and projects without constant user prompting. 

Here’s what each solution offers.

• Slackbot is Slack’s native assistant for Business+ and Enterprise+ subscribers. It can summarize messages and calendars, surface recommendations, correlate data across dashboards, and draft documents. It also plugs directly into Salesforce and Atlassian Confluence, anchoring AI inside tools teams already use.
• Claude CoWork, by contrast, is a research preview for Claude Max subscribers ($100–$200 a month) only available on macOS. It lets Claude work directly with local files—creating new documents, modifying existing ones, and automating tasks. It extends the developer-centric Claude Code into everyday knowledge work. 

Together, these tools position Anthropic less as a standalone AI product  and more as a tool encountered during daily tasks. That enables it to influence productivity and increase user familiarity through repeated use.

Implications for brands: Agentic AI embedded inside everyday tools changes how clients behave—and what they expect. 

When assistants surface answers, options, and next steps inside Slack or a document, people start to expect immediate help where they are working, which speeds up decisions. 

That shift in behavior forces brands to focus less on persuasion and more on usefulness by ensuring clean data, clear positioning, and systems that plug directly into agentic enterprise workflows.