TechCrunch: Salesforce gives Slack a 30-feature AI overhaul

Salesforce announced 30 new AI features for Slack on March 31, 2026, at an event in San Francisco. The centerpiece is the general availability of a rebuilt Slackbot, which the company is positioning as a personal AI work agent for enterprise users. The underlying model is Anthropic’s Claude.

What was announced

The rebuilt Slackbot introduces several distinct capabilities rolling out in phases over the coming months. Reusable AI skills allow users to define specific task templates — such as preparing a meeting brief or drafting a customer follow-up — that Slackbot can apply across different channels and contexts. A built-in library of skills ships with the product; teams can also build custom ones.

Slackbot now functions as a Model Context Protocol (MCP) client, which means it can connect to external services including Salesforce’s Agentforce platform and route tasks to other AI agents without a human intermediary. Meeting intelligence is also new: Slackbot listens to Zoom, Google Meet, and Slack Huddles calls, generates summaries and action items when a call ends, and logs those items directly into Salesforce CRM.

The update also embeds a CRM interface inside Slack. Slackbot reads channel conversations, identifies deal mentions and new contacts, and updates records automatically. An additional capability allows the agent to monitor the user’s desktop, calendar, and communications to surface suggestions and draft follow-ups outside the Slack window itself.

Why it matters for product managers

The announcement represents a structural shift in how Salesforce is positioning Slack. Previously a messaging tool with AI features added on top, Slack is now being built as an orchestration layer for enterprise AI agents — a place where agents from different systems receive work, report back, and coordinate. For PMs at companies running Salesforce CRM, the practical implication is that meeting notes, deal updates, and task routing may happen automatically without switching between tools.

More broadly, this is a signal worth tracking for any PM building productivity or workflow products. The bet Salesforce is making is that enterprise users will accept an AI that operates proactively — watching meetings, monitoring desktops, updating records without being asked. How users respond to that degree of AI presence in daily work is an open product question that will generate real behavioral data over the next several months.