Slack’s New AI Slackbot Is Less About Chat and More About Running the Workplace

Slack bills itself as “where work happens.” Lately, the Salesforce-owned company has been working hard to make that slogan more literal, rolling out features designed not just to speed up how work gets done, but also to keep workers in the app for longer periods. On Tuesday, Slack reinforced that positioning with the general availability launch of its new Slackbot, an AI-powered agent pitched as a personal work assistant.

Its release shouldn’t be surprising, as its arrival was signaled last October. The Slackbot is part of Salesforce’s push to turn the team communication app into the front door for its vision of the “agentic enterprise.” It’s built into Slack and can be used to provide context for work demands, parsing through not only chat histories and channels, but also authorized third-party apps, including Salesforce.

Although Slackbot is now available via a phased rollout, it’s exclusive to Business+ and Enterprise+ customers at no additional charge.

Your Dedicated AI Work Assistant

“We really want to help every single business become an agentic enterprise where humans and agents work side by side, and AI amplifies what people can do rather than replacing them,” Rob Seaman, Slack’s chief product officer and interim CEO, states at a press conference.

“Slack was built from day one to help people get work done through conversations, and increasingly, our interactions with technology are happening through natural language or conversations,” he adds. “We think Slack is the natural place for humans and agents to come together as teammates, and where agents can understand your context and act on your behalf.”

With every major platform offering an AI assistant, what makes Slackbot unique? According to Slack, it addresses two core problems: It knows where users work and also the context of what they do. The Slackbot has no setup cost and promises to immediately understand the worker, their job, and the organization. It has incredible enterprise awareness thanks to its access to an individual’s files, documents, and calendar, along with their tone, their boss’s tone, and other important work context.

“That’s when an AI tool stops being a tool and actually really becomes a teammate,” Ryan Gavin, Slack’s chief marketing officer, attests. Slackbot “is a deeply personal agent…It is your agent. It is not for your teams. It’s your personal productivity assistant, and it’s sitting there right natively in Slack.”

As a demonstration of the agent’s potential, he describes how he asks Slackbot each week where his attention is most needed. Under normal circumstances, that would mean leadership meetings, one-on-ones, and manually scanning Slack channels for signals. Gavin says Slackbot already has that context, compiling priorities without needing to be told who is on his team or what to look for.

Amy Bauer, Slack’s Slackbot product lead, offers a more business-oriented use case, showing how Slackbot can help assemble a summary of a feature pilot and inform plans for broader access. The challenge, she notes, is that the underlying data is scattered: user feedback lives in a busy Slack channel, while adoption and usage metrics sit in external tools.

Slackbot can parse that high-volume feedback to surface key themes and sentiment, while also providing appropriate citations. It can then compare those qualitative insights with product usage data to identify correlations. And because the agent supports all file types, it can interpret files such as dashboard screenshots and graphs; however, it’s currently limited to text-based outputs.

The findings can then be exported into a Slack Canvas for further refinement, either by the team or via the Slackbot, before they’re shared with executives. Finally, users can instruct Slackbot to help coordinate meetings with stakeholders. While the agent can’t yet create calendar events, it can draft agendas and surface links to relevant content, handling much of the prep work around the meeting.

“You can imagine how long this would have taken an employee before, not only do all that research, composite all those answers, and then turn it back,” Gavin points out. “This is multiple employees doing multiple hours of work on what is otherwise, ideally, a simple question to get answered…quickly.”

Seaman chimes in to say that Slackbot has tool-calling capabilities to create a Slack Canvas within Slack. And he reveals there are plans to add support for third parties as well, not only to create Canvases in Slack, but also to “actually be creating documents in other systems, creating tickets in other systems, etc.”

Eliminating the Work of Work

It’s little surprise that Salesforce is among the first to deploy the new Slackbot. Slack’s parent workforce is often the first to pilot these AI features. Early results suggest the company is seeing meaningful gains: according to Gavin, it has become the fastest-adopted product in Salesforce’s history, with teams reporting savings of up to five hours a week and individual employees up to 20 hours. Because of this efficiency, workers can reinvest that time in activities that drive business, increase customer success, and foster product innovation.

Something that surprised Slack through the Slackbot pilot was that Salesforce employees would quickly create and share their favorite and most effective prompts with each other, resulting in a viral, organic spread of best practices across the entire organization. This led Slack to build a shareable prompt feature, allowing users to exchange prompts, embed them in shared messages, and make them easy for teams to discover, copy, and use.

“It happens pretty seamlessly…because it’s intuitive. There’s no training, and it’s happening right where [you] work. [You’re] not having to go to yet another tool,” Gavin explains. Bauer adds that more than three-quarters of Salesforce employees discovered the Slackbot through social sharing. “So many users were discovering Slackbot because they were finding value and then sharing it with somebody that they knew,” she states. “Making that really intuitive within the product where you could share either a prompt or share a response, and then have the prompt be embedded in that response, and then run it in your own user context, is something we’re bringing forth.”

Salesforce isn’t the only one that had early access to this tool—consulting firm Slalom, digital paper tablet maker reMarkable, and Mr. Beast’s Beast Industries are among those now using the Slackbot.

“Their employees are engaging with Slackbot almost instantly because it’s right in the flow of work. It’s right there where they’re doing all their other work inside of Slack. It’s not another tool to learn. It works instantly, and because it knows them, they trust it to work on their behalf. And the productivity gains that they’re seeing are real—90 minutes a day,” Gavin boasts.

He emphasizes that this is only the start for the Slackbot, promising that it will become a “super agent” that deploys across entire organizations, powered by Salesforce’s Agentforce platform. “Slackbot will become your super agent that can know which agent I can pull in at the right time, at the right moment, to help me get the job done.”

That evolution further illustrates how Salesforce envisions Slack being the agentic operating system for the enterprise. If the communication platform is to be the hub of the workplace, eliminating the need for context switching and enabling the management of an army of digital labor are critical. It’s likely that Salesforce views Slackbot as being the trusted chief of staff or lieutenant for every human worker—one bot to control the others, perhaps?

Ensuring a Trustworthy Assistant

Trust surfaced repeatedly during Salesforce’s press conference, and this time it carried more weight than in past AI announcements. Slack’s Slackbot is being pitched as a personal assistant for workers, not a background feature or experimental add-on. If it fails to deliver on that promise, the reputational risk is higher. Users need confidence that the agent will act reliably, respect context, and follow through on its claims. Of all the AI products Salesforce has introduced so far, Slackbot may be the most consequential to get right.

Think about it: If you’re asking Slackbot to handle sensitive information, it’s vital that its responses are accurate, free of hallucinations, and in line with security guardrails and company policies.

Seaman explains that Slackbot uses Retrieval-Augmented Generation (RAG) to provide responses, but only draws from data that the user has access to. “The first thing it does is actually go and search Slack using your user credentials. That means it can access any public channels that are available in your Slack workspace, any private channels that you participate in, and your direct messages because it’s using your user ID,” he remarks. “It’s not doing training or necessarily data storage. It’s performing RAG.”

The company stresses that this agent is built on the same trust foundation as Slack, with respect for roles, permissions, and access controls. It will only surface information that a user has permission to see, and all interactions will remain private.

Slackbot and Developers

For developers, the new Slackbot represents a significant expansion of integration and extension possibilities within Slack. Seaman announces that developers can now build Enterprise Search connectors to connect Slackbot to third-party enterprise systems. In addition, in what is described as the inaugural third-party tool call, Slack’s agent can search across all these connectors.

But that’s not all: Seaman points out that Slackbot will become a Model Context Protocol (MCP) client, allowing developers to hook up their apps directly to Slackbot. “That’s going to be our most exciting area of innovation moving forward,” he declares.

Slack did not provide a timeline for when support for other third-party tool calling capabilities or the transition to the MCP client will be rolled out.

Slackbot ‘Glow Up’

This isn’t the first Slackbot on the platform. Long-time users will recall that there was an earlier incarnation that delivered reminders, sent notifications, and could be configured with custom responses. It would appear through direct messages and could respond to specific commands or automated triggers. However, the program was rudimentary.

“Slackbot got a glow up,” Gavin declares. The previous version was “really just around basic notifications, kind of letting me know something might require my attention. This is an entirely new Slackbot. It has the same wonderful personality that we love to connect with. It’s famous for its taste and design, but it is now a deeply personal system. It’s something that I’m relying on to get work done, not just to let me know something might need my focus.”

Still, what Slack is doing fits squarely into a broader, fast-converging trend. The Slackbot keeps the team communication app competitive with its peers, namely Microsoft Teams with Copilot, Google Workspace with Gemini, and Atlassian with Rovo.

“This is a seminal moment for Slack,” Seaman proclaims. “Slackbot is one of the most powerful features we have shipped to our customers and users in years…we think it’s one of the best examples of humans and agents working together.”

Featured Image: Slack's logo on display at the company's trade show booth at Dreamforce 2025 in San Francisco, CA. Credit: Ken Yeung