The “ping” that has defined modern work is about to get a lot smarter—and a lot more invasive. Salesforce just announced a massive overhaul of Slack, injecting 30 new AI features that transform the platform from a simple messaging tool into a fully autonomous “agentic” ecosystem. We aren’t just talking about better search; we’re talking about a Slackbot that can monitor your desktop, transcribe your meetings, and hire its own “sub-agents” to finish your budget reports.
| Attribute | Details |
| :— | :— |
| Difficulty | Intermediate (Requires workflow setup) |
| Time Required | 15–30 minutes to configure custom skills |
| Tools Needed | Slack Enterprise, Salesforce Agentforce, MCP-compatible apps |
The Why: Ending the “Where Was That?” Era
Most professionals spend 20% of their day just looking for information trapped in old threads or disparate apps. Salesforce is betting that you shouldn’t have to leave your chat window to be productive. By turning Slackbot into an orchestration layer, they are solving the “context switching” tax. Instead of you jumping between a spreadsheet, a calendar, and a CRM, the AI pulls the data to the conversation. If this works, Slack moves from being a distraction to being the actual engine where work happens. This shift mirrors how Microsoft shifts from chatbots to agentic AI with Copilot Coworker, turning assistants into active participants in the digital office.
Step-by-Step: How to Put the New Slack AI to Work
The update rolls out in the coming months. Here is how to hit the ground running when the “AI Makeover” hits your workspace.
- Define Your Reusable “AI-Skills”: Navigate to the new Skill Library. Rather than prompting from scratch every time, create a template for recurring tasks like “Weekly Pipeline Summary” or “Onboarding Checklist.”
- Activate Model Context Protocol (MCP): Connect Slackbot to your external tech stack. Because Slack now acts as an MCP client, you can authorize it to pull live data from non-Salesforce tools to provide a unified view of your projects. Companies are already using Model Context Protocol (MCP) to eliminate data blindness, connecting AI agents to unstructured data for better insights.
- Trigger Agent Orchestration: Use a natural language command like “Organize the Q4 launch.” Watch as Slackbot routes tasks to Agentforce (Salesforce’s agent builder) to identify the right stakeholders and automatically draft meeting invites. This reflects a broader trend where Alibaba’s new Agentic AI framework moves beyond chatbots to automate workflows in Slack and other messengers.
- Set Desktop Privacy Permissions: Go to Settings > Privacy. The new Slackbot can now monitor your desktop activity to “learn” your habits. Audit these permissions immediately to ensure the bot is only watching the apps you want it to see.
- Audit the “Zoned Out” Recaps: Use the transcription tool during Huddles. If you join a meeting late, ask Slackbot for a “Catch Up” to see a bulleted list of what you missed and which action items were tagged to your name. You can also learn how to use Zoom AI Companion to automate meeting summaries to see how competitors are handling similar transcription needs.
💡 Pro-Tip: Use “Skill Chaining.” You can set a custom skill to trigger another. For example, have a “Meeting Summary” skill automatically trigger a “Jira Ticket Creator” skill based on the action items identified in the transcript. This eliminates the manual data entry that usually follows a brainstorm.
The Buyer’s Perspective: Slack vs. Teams vs. Everything Else
Salesforce is clearly feeling the heat from Microsoft Copilot. While Microsoft has a tight grip on the Office suite, Salesforce is leaning into its “open” ecosystem. By making Slackbot an MCP client, they are allowing it to play nice with a wider variety of third-party AI agents than Teams currently allows.
However, the “desktop monitoring” feature is a double-edged sword. While Salesforce promises privacy, many enterprise users may find an AI that watches their “habits” to be a step too far toward bossware. Users who are wary of this may want to explore how Firefox allows you to disable generative AI features to manage privacy more aggressively. The value proposition is clear: extreme efficiency. But the cost is a total surrender of your digital workspace to the algorithm. If you are already deep in the Salesforce/Agentforce ecosystem, this update is a massive force multiplier. If you’re a small shop using Slack for casual watercooler talk, these features might feel like overkill.
FAQ
Can Slackbot actually do work in other apps?
Yes. Through its connection to Agentforce and the Model Context Protocol, it can trigger actions in external tools—like updating a CRM record or drafting an email in a connected client—without you leaving the Slack interface.
Does the AI “watch” me all day?
Technically, it can monitor desktop activity (deals, calendars, habits) to provide context for its suggestions. However, these are opt-in permissions that can be restricted at the admin or user level.
What happens to my data?
Salesforce maintains that these agents operate within a “Trust Layer,” meaning your proprietary company data isn’t used to train the global LLMs used by other companies.
Ethical Note/Limitation: While Slackbot can now summarize and suggest, it cannot yet account for the nuance of office politics or the “unspoken” context of human relationships that often dictate how work actually gets done.