Salesforce finally turned Slackbot into something useful — an AI agent that actually works

Salesforce just dropped a major update to Slackbot, and for once, it’s not just a rebranding exercise. They’ve completely rebuilt the thing — from a glorified reminder bot into what they’re calling an “AI agent” that can search your company’s data, draft documents, and actually take actions on your behalf.

Parker Harris, Salesforce co-founder and Slack’s CTO, didn’t mince words about the upgrade: “The old Slackbot was, you know, a little tricycle, and the new Slackbot is like, you know, a Porsche.” I appreciate the honesty. The old Slackbot was barely useful — it’d remind you to add someone to a doc or suggest archiving a channel. That’s it. Now it’s running on an LLM with real search capabilities across Salesforce records, Google Drive, calendar, and years of Slack conversations.

It’s generally available now for Business+ and Enterprise+ customers, which means most organizations using Slack at scale can get their hands on it. And Salesforce is betting big that this positions Slack at the center of the “agentic AI” trend — where software agents work alongside humans rather than just being passive tools.

Why Anthropic’s Claude is under the hood (and why that might change)

The new Slackbot runs on Claude, Anthropic’s model. Harris was upfront about why: Slack’s commercial service has FedRAMP Moderate certification for U.S. federal customers, and Anthropic was “the only provider that could give us a compliant LLM” when they started building. That’s a practical constraint, not a philosophical choice.

But exclusivity won’t last. Harris said they’re adding support for other providers this year. “We have a great relationship with Google. Gemini is incredible — performance is great, cost is great. So we’re going to use Gemini for some things.” OpenAI is also on the table. He echoed Marc Benioff’s view that LLMs are becoming commodities: “I call them CPUs.”

On the training data question — which matters a lot for enterprise customers — Harris was blunt: Salesforce does not train any models on customer data. “Models don’t have any sort of security,” he explained. “If we trained it on some confidential conversation that you and I have, I don’t want Carolyn to know — if I train it into the LLM, there is no way for me to say you get to see the answer, but Carolyn doesn’t.” That’s the right answer for any serious business tool.

80,000 employees tested it internally — and the numbers are impressive

Salesforce rolled Slackbot out to all 80,000 employees months before the public launch. According to Ryan Gavin, Slack’s CMO, “It’s the fastest adopted product in Salesforce history.” That’s not just marketing fluff — the data backs it up.

Two-thirds of Salesforce employees have tried it, and 80% of those keep using it regularly. Internal satisfaction hit 96%, which is the highest for any AI feature Slack has shipped. Employees report saving between two and 20 hours per week. That’s a huge range, but even the low end is meaningful.

What’s more interesting is how adoption happened: organically. Within five days, employees created a shared Canvas called “The Most Stealable Slackbot Prompts” that now has over 250 prompts. Kate Crotty, a principal UX researcher, found that 73% of internal adoption came from social sharing, not top-down mandates. “Everybody is there to help each other learn and communicate hacks,” she said. That’s the kind of organic growth you can’t buy.

What it actually does in practice

During a demo, Amy Bauer, Slack’s product experience designer, showed Slackbot synthesizing customer feedback from a pilot program, uploading a usage dashboard image, and correlating the data — all in one conversation. It’s the kind of cross-source analysis that used to require multiple tools and a lot of manual work.

But here’s where I get a bit skeptical. Salesforce is positioning this as the “front door to the agentic enterprise,” and Harris claimed it’s “not just another copilot or AI assistant.” That’s a bold statement when Microsoft is pushing Copilot into every corner of Office 365 and Google is doing the same with Gemini in Workspace.

The reality is that Slackbot is now a strong contender, but it’s still tied to the Salesforce ecosystem. If your company lives in Slack and Salesforce, this is a no-brainer. If you’re on Microsoft Teams or Google Chat, the integration story is less clear. Slack has always been the independent player, and that independence is both a strength and a weakness.

The bottom line

Salesforce has done something genuinely useful here — they took a forgettable feature and turned it into something that might actually change how people work. The internal adoption numbers are real, and the product seems to deliver on its promises. But the workplace AI battle is just heating up, and Microsoft and Google aren’t sitting still. Slackbot’s success will depend on whether it can integrate deeply enough with the tools people already use, not just the ones Salesforce sells.