Salesforce’s new Slackbot is a Porsche, not a tricycle — and it runs on Claude

Salesforce just dropped a completely rebuilt Slackbot, and it’s not your grandma’s notification bot anymore. This thing is now a full-blown AI agent that can search your enterprise data, draft documents, and take actions on your behalf. Think of it as the difference between a tricycle and a Porsche — those are the words of Slack CTO and Salesforce co-founder Parker Harris, not mine.

The old Slackbot was fine for reminding you to add someone to a doc or archive a channel. But the new one? It’s built around an LLM and a serious search engine that can pull from Salesforce records, Google Drive, calendar data, and years of Slack conversations. It’s generally available now for Business+ and Enterprise+ customers.

Why Anthropic’s Claude is under the hood

Here’s where it gets interesting. The new Slackbot runs on Claude, Anthropic’s model. That’s not just a random choice — Slack needed FedRAMP Moderate certification for government customers, and Harris told me Anthropic was the only provider that could offer a compliant LLM when they started building. But that exclusivity won’t last. He said they’re planning to support additional providers this year, including Google’s Gemini and possibly OpenAI. Harris echoed Marc Benioff’s view that LLMs are becoming commodities — he called them “CPUs.”

And no, Salesforce isn’t training models on your data. Harris was blunt about that: “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.”

The internal numbers are impressive

Salesforce tested this thing internally with all 80,000 employees. According to Slack CMO Ryan Gavin, it’s the fastest adopted product in company history. Two-thirds of employees tried it, 80% of those kept using it regularly, and internal satisfaction hit 96%. Employees report saving between two and 20 hours per week. That’s a wide range, but even the low end is meaningful.

What’s more telling is how adoption happened. Kate Crotty, a principal UX researcher, found that 73% of internal adoption came from social sharing, not top-down mandates. Employees built a shared Canvas called “The Most Stealable Slackbot Prompts” — it now has over 250 prompts, all added organically. That’s the kind of organic adoption you can’t buy.

The bigger picture: Slack as the agentic front door

Harris positioned Slackbot as “the front door to the agentic enterprise, powered by Salesforce.” That’s a bold claim, especially when Microsoft and Google are pushing their own AI assistants hard. But Slack has a unique advantage: it’s where people already live in their workday. If Slackbot can actually pull data from across Salesforce, Google Drive, and other tools without making you hop between apps, that’s genuinely useful.

During a demo, product experience designer Amy Bauer showed Slackbot analyzing customer feedback from a pilot program, correlating it with a usage dashboard image, and producing executive-ready insights. That’s the kind of cross-source synthesis that’s been promised for years but rarely delivered cleanly.

My take

I’ve seen plenty of AI assistants that sound great in press releases but fall apart in practice. The internal adoption data here is stronger than most, but the real test is whether it works as well for external customers as it did for Salesforce employees. The model flexibility is smart — locking into one LLM provider long-term would be a mistake, and Harris seems to get that.

Also, keeping the Slackbot name despite the full rebuild is the right call. People know what Slackbot is, and rebranding would just confuse everyone. Sometimes the boring choice is the smart one.

If you’re on Business+ or Enterprise+ Slack, go try it. If you’re not, you might find yourself asking your IT team why you’re still stuck with the tricycle.