Salesforce Inc. said today that its collaboration tool Slack is evolving to provide developers with new tools to connect artificial intelligence applications securely to conversational data.

New additions to the platform for developers include a real-time search application programming interface, a Model Context Protocol server and enhanced developer tools for building, configuring and administering agentic AI apps.

As enterprise businesses push forward, building AI agents, they need to connect them to the everyday workflows of employees and customers. Agentic AI runs on data, and access to it is fundamental to its operation since data provides context and understanding. According to Salesforce, conversational data provides some of the richest context for agents to deliver personalized assistance.

The new real-time search API and Model Context Protocol server allow developers to connect securely to Slack’s collaborative space and deliver data securely.

The RTS API provides real-time access to conversational data on demand, enabling AI applications to retrieve current discussions, files and channels without needing to capture everything and sift through it. That relieves developers of the burden of making bulk data transfers and storing large amounts of potentially sensitive data locally to send to AI models.

Kurtis Kemple, head of Slack developer relations at Salesforce, told SiliconANGLE in an interview that the new approach reduces the time and energy developers spend writing complex code or making multiple API calls to accomplish a task.

“A workflow between a team of designers, some front-end engineers and agents from Vercel and Figma — we’re at 30 or 40 messages now,” Kemple said. “You’re going to have to make 40 API calls and grab 40 messages. Should it be like that every time?”

He explained that the new RTS capability streamlines this process, lowering the number of API calls required while serving up only the conversational data developers need to feed into AI applications. Not only does that reduce coding complexity, but it also lowers overhead and cost.

The MCP server acts as a standard for connecting large language models, AI apps and agents to contextual information. It provides a unified communication layer between any LLM and Slack, reducing the overall coding work developers need to connect data sources and tools.

Developed and released as open-source by Anthropic PBC, MCP reduces fragmented integrations with a single consistent protocol that fits user specifications and can be tuned to their needs. With MCP, developers don’t need to define manually every possible task an agent can perform or implement complex connectors for each LLM.

Additionally, the Slack platform is standardizing how developers build and connect dynamic experiences with Slack Work Objects. Work Objects standardize how third-party data is displayed in Slack, such as detailed information, images and documents. It also allows teams to connect to third parties, such as Asana Inc., and take action and view results without leaving the platform.

New agentic developer tools include prebuilt Block Kit Tables and command-line interface capability for Bolt apps. Block Kit Tables are a native way for developers to display structured, tabular data within messages, including columns and rows. Providing them prebuilt allows developers to quickly fill them up with data without having to define message layouts. The new CLI feature enables developers and administrators to efficiently configure and launch applications using scripts and automation tools.

“Because if you want to build AI — or even if you don’t, if you’re building anything — it all starts with, ‘I gotta have an app,’” Kemple said. “A developer has to build that.”

Major companies, including Anthropic, Google LLC, Perplexity AI Inc., Writer Inc., Dropbox Inc. and Anysphere Inc.’s Cursor are already building on these tools with new AI Slack apps and agents.

For example, Anthropic’s Claude chatbot can integrate directly into Slack workspaces, allowing teams to talk to the chatbot via direct messages, an AI assistant panel or @mentions in threads. This includes access to web search, enterprise data sources and document analysis. Perplexity is using RTS to integrate AI-driven web search capabilities with insights from internal conversations, providing real-time research that intelligently addresses user needs.

“We’re thinking about what AI is like and what we need to be able to support three years from now, four years from now: How is this line of business going to change?” said Kemple. “We want to make sure that we’ve got the right place for people to come and do those jobs.”

Images: Slack, Salesforce