Monday morning. You open Slack and last Friday's meeting summary is already there. Your action items are laid out. Your Salesforce CRM has already been updated. This is Slack in April 2026. On March 31st, Salesforce shipped 30+ AI features to Slackbot all at once. It's the most ambitious update since the $27.7 billion acquisition in 2021.
What Is It?
One-liner: The era has arrived where AI agents inside the Slack you use every day listen to meetings, create documents, update your CRM, and connect apps. Slackbot has gone from a simple chatbot to a "digital teammate."
Here's the background. On January 13th this year, Slackbot got its first-ever agent capabilities — drafting emails, scheduling meetings, triaging inboxes. Then in under three months, they shipped 30+ more features on top of that. Salesforce CTO Parker Harris put it this way: "Slackbot is an always-on, superintelligent teammate."
Some numbers that are already showing up in the real world:
Slack EVP Rob Seaman said it best: "The ceiling on what Slackbot can do is essentially unlimited." And two months in, it's on track to become the fastest-adopted feature in Salesforce history.
You don't need to go through all 30+ features. Here, I've picked the 6 core ones worth using right now.
What Changes?
Until now, Slack was "where you talk." Now Slack wants to be "where work gets done." The old pattern: discuss in Slack, execute in another app, come back to Slack to report. This update is designed to eliminate that gap.
| Old Slack | AI Slack (Apr 2026+) | |
|---|---|---|
| Post-meeting workflow | Manual notes, manual CRM updates | Auto-summary + instant CRM sync |
| Recurring tasks | Write a new prompt every time | Define once with AI Skills, auto-applied |
| External app integrations | Open each app separately | Call 2,600+ apps from one chat window via MCP |
| Customer management (small teams) | Spreadsheets or a separate CRM purchase | Native CRM inside Slack, auto-managed |
| Research | Search engine → gather info → manual organization | Deep Research does a thorough analysis in ~4 minutes |
| What AI can know | Only the current conversation | Remembers channel history, files, and your work habits |
The key is "context advantage." Slack holds years of accumulated conversations, decisions, and files. Other AI tools don't have this context — but Slackbot does. Seaman described it as "competitor aware, customer obsessed."
The 6 Core Features, Up Close
1. AI Skills — Standardizing Recurring Work
The most basic — and most powerful — feature in this update. An AI Skill is a reusable instruction set. Say "build me an event budget" and Slackbot gathers relevant info from channels, pulls in connected apps and data, drafts an execution plan, and books a meeting based on participants' job titles.
One person's Skill becomes the team's standard. Slackbot automatically applies a Skill when a prompt matches it — so the quality stays consistent even if users don't know the Skill exists.
2. Auto Meeting Summaries + CRM Integration
Zoom, Google Meet, any video conferencing tool — the Slack Desktop app listens to meetings through local audio, summarizes the discussion, and extracts action items. The difference from existing meeting tools is clear: it doesn't stop at the summary. It goes all the way to execution. The moment a meeting ends, deal updates and next steps are automatically logged in Salesforce CRM.
3. Desktop Agent
Slackbot steps outside the Slack app and works alongside you on your desktop. Drawing on your transactions, calendar, conversation history, and work habits, it makes context-aware suggestions and auto-drafts follow-ups. Select a contract on your screen and say "extract the pricing table and update the vendor record" — it handles it.
Privacy Concerns?
Seaman emphasized that "every feature is user opt-in." Meeting recording only starts when you explicitly ask for it, and the desktop agent only reads context when you capture and share your screen. There are no plans to expose personal memory data to admins, and you can tell it to "forget" anything at any time.
4. MCP Client — A Hub for Every App
Slackbot is now a Model Context Protocol (MCP) client. What that means: it can connect to external services and tools to execute tasks. Not just Salesforce's Agentforce — but 2,600 apps on the Slack Marketplace and 6,000 on Salesforce AppExchange, all callable through a single conversation.
"Seaman said he's going all-in on MCP." Create a Google Slides deck, draft a Google Doc, update a Linear issue, deploy on Vercel — users don't need to know what system is running under the hood.
5. Native CRM (For Small Teams)
For startups and small teams running on Slack without a dedicated CRM. Slackbot reads your channel conversations and automatically manages deals, contacts, and call notes. When you scale up, all your data is already connected to Salesforce — no migration needed.
Seaman is refreshingly blunt about it: "There's always a moment when a growing company needs a CRM. If you're already inside Slack at that point, there's no reason to go buy another tool." It's also Salesforce's direct shot at the trend WSJ reported of startups "vibe-coding their own CRMs."
6. Deep Research + Voice Input + Memory
Deep Research is a mode that runs multi-step, in-depth analysis over roughly 4 minutes. It's built for depth over speed — so much so that Salesforce deliberately skipped demoing it in the keynote. It searches external web news, then automatically cross-references that with internal Slack conversations to surface "what this means for our business."
Voice input supports TTS/STT, and full speech-to-speech conversation is in the works. The memory feature learns your preferences, workflows, and habits over time to deliver increasingly accurate responses.
Getting Started
- Check your plan
Slackbot AI is available on Business+ ($12.50/month/user) or Enterprise Grid. Starting in April, Free and Pro users can try it with limited access. Enable it with a single toggle in Admin Settings. - Start a conversation with Slackbot
Go to the Slackbot DM and ask in plain language. Something like: "Summarize the campaign budget discussions in #marketing from last week." It answers based on Slack's internal data. - Build your first AI Skill
Pick one task you repeat every week and turn it into a Skill. A great starting point: "Every Monday, summarize the pipeline status in #sales and post it." Save it to Canvas to share it with your team. - Turn on meeting notes
Before your next video call, ask Slackbot to "take notes for this meeting." As long as the Slack Desktop app is running, it works with Zoom, Meet, or any other tool. The summary appears as soon as the meeting wraps. - Check your connected MCP apps
In Admin > Apps, review the list of Slack apps you've already installed. Slackbot can now call these apps through MCP in conversation. Major apps like Cursor, Vercel, Linear, and DocuSign are already supported.
Heads Up
Slackbot AI's reasoning engine runs on Anthropic Claude. Slack absorbs the cost, but that means there may be usage limits. There's also a risk Seaman himself acknowledged: "With 30 features, there's a real danger of losing the simplicity that made Slack great. That's what keeps us up at night." You don't have to turn everything on — pick what your team actually needs.
