In short: Salesforce unveiled more than 30 new AI capabilities for Slackbot on 31 March 2026, the most sweeping overhaul of the platform since its $27.7 billion acquisition in 2021. The update transforms Slackbot from a conversational assistant into an agentic system that can transcribe meetings across any video platform, monitor users’ desktop activity, execute tasks through third-party tools via the Model Context Protocol, and serve as a lightweight CRM. The new features run on Anthropic’s Claude, are live now for Business+ and Enterprise+ subscribers, and will roll out in limited form to free and Pro users from April. From summer 2026, Slack will be automatically bundled with every new Salesforce customer account.

Salesforce has given Slack its most ambitious overhaul since buying it for $27.7 billion four years ago. At an event in San Francisco on 31 March 2026, CEO Marc Benioff unveiled more than 30 new AI-powered capabilities for Slackbot, the platform’s built-in assistant, in what the company described as a transformation from a messaging tool into what it calls an “agentic operating system” ,a single surface where workers interact with AI agents, enterprise applications, and one another. The announcement builds on a January 2026 update that first gave Slackbot the ability to draft emails, schedule meetings, and search inboxes; the March additions are considerably more expansive.

What the 30 features actually do

The centrepiece of the update is reusable AI skills. Users can define a specific workflow once, “summarise this campaign brief“, “generate a budget plan for this event“, and save it as a named skill. Slackbot then learns to recognise when that task is being attempted and offers to run the skill automatically, pulling together relevant information from connected channels, applications, and data sources without the user needing to configure the process each time. The expectation that AI handles repetitive cognitive labour without requiring users to configure pipelines manually has become a baseline demand in enterprise software, and reusable skills is Slack’s answer to it.

Meeting intelligence lets Slackbot listen to calls on Zoom, Google Meet, or Slack Huddles by tapping into a user’s desktop audio through the desktop application. It does not merely transcribe: it identifies decisions made, assigns action items to the relevant participants, and delivers a structured summary automatically when the call ends. The desktop agent extends Slackbot’s reach beyond the Slack application itself, monitoring screen activity and drawing on a user’s deals, conversations, calendar, and habits to surface suggestions and draft follow-ups proactively. This is the feature with the most significant privacy surface: Slack is, in effect, asking users to grant an AI agent persistent visibility into their computer.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

Slack is also launching a native CRM for small businesses, built directly into the chat interface. Slackbot reads channels, identifies when a deal is mentioned or a new contact introduced, and updates those records automatically. Salesforce says companies that outgrow the lightweight version can migrate to full Salesforce CRM without rebuilding their data. Finally, Slackbot now functions as an MCP client, a Model Context Protocol client, meaning it can connect to and coordinate with any external service that registers an MCP server through Slack’s manifest. That list currently includes Agentforce (Salesforce’s own AI agent platform launched in 2024), Google Workspace, Microsoft 365, Notion, Workday, ServiceNow, and more than 6,000 other applications in the Salesforce ecosystem. The shift toward agent orchestration, where a single interface coordinates multiple AI systems across different enterprise applications, is defining the competitive landscape for enterprise software in 2026 more than any other technical development.

Claude, FedRAMP, and why Anthropic got the contract

Under the hood, all of Slackbot’s new capabilities run on Anthropic’s Claude. Salesforce says it chose Claude partly because Anthropic was the only AI provider meeting FedRAMP Moderate certification requirements at the time Slack began designing the new system, a prerequisite for selling into regulated industries including government, healthcare, and financial services. Anthropic’s trajectory as an enterprise AI provider has accelerated considerably since that decision was made, and the Slack partnership is one of the most visible deployments of Claude in a major productivity platform. Salesforce reports that Slackbot already has nearly one million weekly active users, a figure that is about to grow substantially given the pricing and bundling changes that accompany the announcement.

A leadership backdrop that tells its own story

The March 31 event was hosted by Marc Benioff directly, a signal of how central the Slack transformation is to Salesforce’s current narrative. The more revealing context, however, is what happened three months earlier: in December 2025, Denise Dresser, Slack’s CEO since 2023 and a Salesforce veteran of 14 years, left the company to become OpenAI’s first chief revenue officer. Her successor as Slack CEO is Rob Seaman, the platform’s former chief product officer, who holds the role on an interim basis. That the executive who built Slack’s commercial strategy chose to leave for OpenAI, the company whose products represent perhaps the biggest threat to any single-platform AI assistant strategy, is a detail that Salesforce has had to absorb while executing its most ambitious product push in years.

The Microsoft problem

The 30-feature release is also, unmistakably, a competitive response. Microsoft has been building toward a deeply integrated AI layer across its 365 productivity stack, embedding Copilot into Teams, Word, Excel, Outlook, and PowerPoint, creating a coherent agent experience for organisations already inside the Microsoft ecosystem. Slack’s counterargument, made explicit in the “agentic operating system” framing, is that a communication-first interface with broad enterprise integration is a better home for AI agents than a document-first one, and that MCP connectivity means Slackbot does not require an all-in bet on any single platform. The argument is more credible with 30 new capabilities than it was six months ago.

Pricing, availability, and the bundling play

Slackbot is already included in Business+ and Enterprise+ plans at no additional consumption cost. Starting in April 2026, a more limited version will become available to users on Slack’s free and Pro plans. The more strategically significant change comes this summer: from that point, every new Salesforce customer will have Slack automatically provisioned and AI-enabled from day one. That bundling decision removes the question of whether enterprise buyers will pay for a separate AI layer, they will not need to, because it will arrive with the CRM they already purchased.

The broader enterprise expectation in 2026 is that AI tools meet the same security and compliance standards as the infrastructure they sit alongside, and Slack’s decision to anchor its new system on a FedRAMP-certified model addresses that expectation directly. Whether that is enough to shift purchasing decisions in accounts that have already standardised on Microsoft 365 is a different question, one that the next few quarters will begin to answer. What the March 31 event established is that Salesforce has decided Slackbot is the most important product it is building, and it is acting accordingly.