Slack gets even more access to your work chats to try and build better agents (www.techradar.com)

Salesforce announced major Slack upgrades designed to feed agentic AI with the raw conversational context it needs: a real-time search (RTS) API that unlocks unstructured chats, files and channels for use by AI tools; an MCP (Model Context Protocol) server based on Anthropic’s open standard to standardize integrations across enterprise systems; and an improved developer CLI to speed building AI apps and agents inside Slack. Together these components let AI access live conversational signals without first exporting and restructuring data, while the MCP server aims to make integrations between models and back-end systems more portable and consistent. For the AI/ML community this matters because conversational data — long locked in ephemeral, unstructured messages — is a rich contextual signal that can significantly improve agent relevance, grounding and multi-step reasoning. Practically, teams can build agents that deliver more accurate, timely answers and workflows by using Slack as both the UI and the context layer, while interoperability via MCP should simplify multi-model, multi-service deployments. The move also raises operational considerations for model training, prompt context windows, data governance and privacy controls since enterprises will need policies and tooling to secure and audit agent access to sensitive conversational content.