Slack said on Thursday that it is now allowing developers to create agentic workflows and artificial intelligence (AI) applications using its conversational data. The foundation of the system is a specialised Model Context Protocol (MCP) server, an application programming interface (API) for real-time search (RTS), and developer-focused tools like Block Kit Tables. This will allow third-party developers and even businesses to create safe, context-aware AI apps and agents that may leverage the platform’s chat data to automate tasks in an enterprise-focused communication and collaboration platform.
The significant platform extension announced by Slack will provide developers with the resources and safe access to its conversational data they need to create advanced AI bots and apps. With this change, Slack will no longer be a communication platform but rather a “agentic work operating system” that allows AI and humans to collaborate easily.
Due of their lack of context, AI assistants in workplace products have always had difficulty. Behind data silos, everything from task-based nuance to conversation histories and thread details is dispersed. By granting access to its conversational data to outside AI app and agent developers, Slack hopes to remove those restrictions, according to a press release.
Slack wants to use an MCP-based server from Anthropic and an RTS API. The API provides real-time access to the platform’s conversational data, which includes open channels, files, and debates. By limiting access to the data to AI agents and prohibiting bulk downloads and storage, the company claims to have prioritised security. Businesses can also set the rules for access permissions.
The MCP server, on the other hand, sets a standard for how LLMs, AI applications, and agents find contextual data and carry out tasks for Slack users. For connecting to data sources and tools, it provides a single communication layer between Slack and the LLm. This spares developers the trouble of having to specify each task that an agent is capable of performing by hand.
Several firms have already started incorporating this option, according to Slack. Claude from Anthropic, for example, may now search across Slack files, threads, and channels to contextualise answers in discussions about work. Google’s Agentspace integration connects its AI agents to Slack data in a similar manner. The integration has also been embraced by Perplexity, Writer, Dropbox, Notion, Cognition Labs, Vercel, and Cursor.
In addition, Slack brought new agentic developer tools, such as prebuilt Block Kit Tables, AI best practices, and updated CLI resources for Bolt apps, which streamline the entire build lifecycle. Work Objects are standardised, rich previews that link third-party data (details, images, documents) directly to conversations.
Both the MCP server and the RTS API are the new features under closed beta, with a general release date of early 2026. These features can now be found in third-party AI agents on the Slack Marketplace.
By the end of October, Slack Work Objects, a tool that will assist in integrating third-party data directly into discussions, will be expanded.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
