Slack Gives Developers Live Access to Team Data for AI Agents
With the Real-Time Search API and MCP server now generally available, AI agents can pull real-time context from Slack without compromising security
Developers looking to ground their AI agents in live workplace data can now tap Slack’s real-time search (RTS) API and its Model Context Protocol (MCP) server. Months after being announced, the two developer tools are now generally available. When incorporated, agents can leverage project conversations, team knowledge, business decisions, and other Slack data to inform their responses.
The company boasts that, in the four months since these features were introduced, more than 50 partners have built Slack-powered bots. Furthermore, agent activity on the team communication app has increased by 25x across both RTS queries and MCP tool calls.
“Slack is the system of engagement where humans and AI agents work together,” Rob Seaman, Salesforce’s executive vice president and Slack general manager, tells The AI Economy. “With the Real-Time Search API, customers can now give agents the ultimate contextual goldmine—the ability to securely tap into the live pulse of their company’s knowledge. By grounding AI in this real-time context, we’re moving to a world where agents act as teammates, solving problems, and executing tasks in the flow of work.”
Both the RTS API and the MCP server can be considered part of the company’s effort to protect customer data from being used to train models or for other third-party purposes. Last year, the Salesforce-owned company revised its terms of service, prohibiting developers from indexing, copying, or permanently storing Slack messages through its API. The availability of the RTS API and the MCP server are two new ways to control data flow, but for the agentic era.
With the platform’s RTS API, apps can securely access Slack data via queries, retrieving only relevant information without storing customer data on external servers. As for the MCP server, it dictates how agents can access Slack data and provides the conversational context needed to ground bot responses and actions.
Slack reveals that customers can find agents using these new developer tools in the Slack Marketplace. They span different categories, from agentic search (Claude, Guru, Manu, and Perplexity) and productivity and task management (Ari, Read AI, Tembo, and Spotter) to workflow automation and content (Alysio GTM AI, Credal AI, Moveworks, and WRITER), and function and use-case-specific bots (Claude Code, Cursor, Nalvin, and Wordsmith).
“We continue to deepen our integration with Slack through our MCP Apps extension and by including the power of Claude Code,” Scott White, Anthropic’s head of product for Claude, writes in a statement. “Customers are increasingly seeing the value of this bi-directional integration, and usage is accelerating as a result: faster answers grounded in real team context, improved workflows, and insights delivered right in Slack, where work is happening.”
Updated on Feb. 17, 2026 at 5:22 p.m. PT: Added statement from Slack GM Rob Seaman.
This is exactly the pattern playing out across the board. Slack with RTS + MCP, and now Google just shipped a Workspace CLI that does the same for Drive, Gmail, Calendar, Sheets. The interesting bit is it auto-generates its entire command surface from Discovery Documents so it never goes stale when the API changes. Wrote about the architecture: https://reading.sh/google-workspace-finally-has-a-cli-and-its-built-for-agents-5f5fe87d0425