What the new Slackbot actually is

For most of Slack's history, Slackbot was just a notifier. That changed when Slack rebuilt it from scratch as your personal agent for work. Now, with the latest update, it works across your third-party agents, surfaces business context, and executes tasks across apps, all from one conversation.

The short version:  Slackbot now draws on your channels, files, conversations, and connected apps to bring the right context into every conversation — so it already knows the project, the people, and the decisions that led to this moment. It's powered by Anthropic's Claude, and the quality of its answers comes down to what Slack calls "context engineering," pulling the right slice of your workspace into the model at the right time.

Here's what that looks like in practice.

Reusable AI skills

This is the feature that changed how we work the most. You define a task once (the inputs, the steps, the output format) and save it as a skill. From there, anyone on the team can trigger it with a single command.

We built one for story research. Before an interview, a writer triggers the skill, and Slackbot pulls together everything discussed about that topic across our channels: past pitches, source conversations, background research, and relevant docs. It builds a briefing in seconds. After the call, a second skill takes the recording and pulls out key quotes, follow-ups, and anything that needs flagging. No more scrubbing through transcripts.

Slack ships a built-in library of skills, but the custom ones are where teams will get the most value. Think "create a budget for this event" or "generate a weekly standup summary from these three channels." You build the workflow once, then it just runs.

MCP and app orchestration

Slackbot now functions as an MCP client (Model Context Protocol), which means it can connect to and coordinate with external tools: Agentforce, Google Workspace, Microsoft 365, Notion, Workday, ServiceNow, and thousands of other apps in the Salesforce ecosystem.

In practice, this means you can ask Slackbot to do something that spans multiple systems and it figures out the routing. A budget request in Slack triggers Slackbot to pull the relevant data, draft a plan, and schedule a meeting with the right people. You don't need to know which tool handles what.  You just describe what you need and Slackbot coordinates across those systems to get it done.

What we noticed after a few weeks

The thing that sticks isn't any single feature. It's the accumulation. A quick "summarize this thread" before jumping into a meeting. A "draft a follow-up based on what we discussed" right after one. A "what did I miss in #editorial today?" when you've been heads-down on something else.

None of that requires setup, training, or a new interface. Slackbot is already inside the tool you're already using, already reading the messages you're already sending. You don't adopt it. You just start asking it things.

Slack says some teams are reporting up to 90 minutes saved per person, per day. From what we've seen over a few weeks, the number feels real. Not because the work disappears, but because the time spent hunting for context and re-explaining things does.

Try it in your workspace

If you're on Slack Business+ or Enterprise+, Slackbot is available now. No setup needed. Just look for it next to your search bar and start a conversation. Starting in April, a limited version is rolling out to free and Pro plans too.

Watch the on-demand event to see live demos and hear how teams like Anthropic are putting it to work.