The AI that knows your work and organization

Every few months, something shifts. What you’re doing in Slack today wasn’t possible three months ago.

In April, Slackbot learned to act. In May, it learned to see. On June 24, it learns you, your team, and your entire stack. And it listens.

This is the story of those three months. And why the last chapter changes what the first two made possible.

April: Slackbot learned to act

The team built it because asking Slackbot to help and then doing the work yourself isn’t help. In April, that changed. Slackbot could now take action: create records in Salesforce, run tasks on a schedule, build skills that did the work inside the systems your team already runs on.

An AI that answers questions is useful. But answering a question about a deal versus actually updating the deal record, logging the activity, and flagging the follow-up, are two very different things. One saves you a few seconds of searching. The other improves your output and moves your work forward.

Raveesh Raina, a sales engineer at Salesforce, built a skill to log his activity after customer calls. He shared the skill and within three weeks, 466 of his colleagues were using it. 73% found it through word of mouth. Forty-three minutes saved, per person, per week.

That’s what it looks like when one person’s best process becomes everyone’s capability.

 

May: Slackbot learned to see

The team built it because the best answer in the world doesn’t help if it lives in someone else’s browser history. In April, Slackbot could act. In May, it could see.

In May, Slackbot could reach outside Slack and bring what it found back into the conversation. Search the web, read a PDF, build a Native Chart from raw numbers, or highlight any text and ask Slackbot to explain it in context. The output didn’t disappear into a separate tool or a file no one would open. It landed right where the work was already happening, ready for the whole team to see and build on.

This is the part that matters for everyone else on the team. The person who did the work didn’t have to translate it into a separate deliverable. The conversation was the deliverable.

  

June 24: Slackbot learns you. Your team. Your stack. And it listens.

You’re walking out of a meeting. An idea lands. You don’t have a free hand.

That’s when you need Slackbot. Voice Actions make that possible. Open Slackbot on your phone and say what needs to happen: log the call, update the record, capture the thought before it disappears. Slackbot already knows your context, so you don’t have to re-explain. You just say it out loud.

That works because Slackbot already knows you. And the longer you use it, the more it knows.

And it doesn’t wait to be asked. Slackbot can act on a schedule, fire when a message lands in a channel, kick off when someone adds a reaction. That’s a different kind of AI than one that shows up when you tag it.

The AI that knows your team, not just you.

April and May were about capability. June goes deeper on something that’s harder to build: context. The team built it because we want AI that knows our work. April and May were about capability.

Most AI has memory now. But there’s a difference between an AI that remembers what you told it and one that knows what your organization knows. Some AI reads the room when you ask it to. Slackbot has been in the room all along. Every other AI learns what you told it: your role, your preferences, the context you typed into a settings panel. Slackbot learns what your team actually did. The decision made in a channel last month. The project thread where the real answer lives. The Salesforce deal context from the call that happened in Slack. You didn’t have to brief it. It was already there.

The more you work in Slack, the sharper it gets. Ask it to draft a follow-up on a deal you haven’t touched in six weeks and it already knows the stakeholders, the sticking points, and the last thing your champion said. Not because you told it. Because that’s where the work happened. That context compounds.

It doesn’t stop with you.

Memory makes Slackbot better the more you use it. Skills spread that to your team.

Think about the last time a colleague figured something out that would have saved you an hour, and you never knew. The perfect way to prep for a renewal call. A prompt that turns a messy data pull into something a VP can read in 30 seconds. That kind of knowledge used to travel by word of mouth, if it traveled at all. With Slackbot there’s a way to share and build on what your colleagues have already figured out. When someone on your team builds a skill, everyone benefits.

Your whole stack, one conversation.

Context is only useful if it can follow you. With MCP, it does. Most teams run on five or more apps. With MCP, Slackbot brings your whole stack into the conversation — Google, Atlassian, Box, Notion, DocuSign, and more. You don’t have to leave to get the answer. You don’t have to re-explain the context. Slackbot reaches where the work is and brings it back to where the conversation is already happening.

That’s the through-line across all of this. Memory makes Slackbot better for you. Skills spread that to your team. MCP brings the rest of your stack into the conversation. Every employee gets an AI that gets smarter not just because they used it, but because everyone around them did too.

And, for the employee who never opened Slackbot because typing felt like one more thing, you can now just say it out loud.

Open Slackbot. Ask it: “What do you know about me?”

That’s where it starts. And it gets better from there.