Slack scored 6.0/10.

The page opens with “All your people and AI agents working together” — a headline that describes a category shift Slack is betting on, not a reason for any specific buyer to stay on the page. The problem is that Slack’s actual visitor population splits hard between two groups: teams already using Slack who came to understand what the AI layer does for them, and buyers evaluating Slack against Microsoft Teams who need a reason to choose it. The headline serves neither. The existing Slack user needs to know what changes for them specifically — what Slackbot does that they can’t do today. The Teams evaluator needs to know why Slack wins on the AI dimension, not just that AI exists here too. One headline cannot hold both audiences, and right now it holds neither. Split the page or pick the buyer who matters more to the business right now, then write the hero for that person.

The page makes four distinct claims across its four sections — Knowledge, People, Process, Platform — and treats them as equal pillars. This is the page trying to own four things simultaneously, which means it owns none of them. The actual differentiating claim buried in the body is Slackbot: an AI agent that knows your team’s history, coordinates across apps, and gets work done inside a single conversation. That claim is genuinely differentiated from what Microsoft Teams is shipping and what standalone AI tools can offer. But it appears in the third paragraph of the page, after a section header, after a stat callout, after the hero. The page’s architecture treats Slackbot as one feature among many. It should be the organizing frame for the entire page — the thing everything else proves.

The testimonials from Anthropic and Vercel are doing the wrong job. Guillermo Rauch says “we see Slack as the natural place to build our agents” — that’s a developer-building-on-Slack quote, not a buyer-using-Slack quote. Kate Jenson says Slack helped Anthropic stay aligned as they scaled — that’s a generic collaboration endorsement that Microsoft Teams could run without changing a word. Neither quote names the specific outcome that Slack’s positioning depends on: that Slackbot or AI in Slack did something no other tool could do. The page has a case study about MrBeast and Slackbot helping create viral videos — that’s a more specific, memorable proof point than either named testimonial — but it appears near the bottom, after the reader has already formed their impression. Move the most specific AI proof to the top; replace the generic endorsements with quotes that name what the AI actually did.

The stat “97 minutes saved weekly with AI in Slack” appears mid-page with a footnote that traces it to an internal pilot analysis. The number is credible and specific, which is good. But the page never tells the buyer what those 97 minutes were spent on before, or what the buyer does with them after. “97 minutes saved” is an input metric; the buyer wants to know the output — what their team ships, closes, or resolves with that time back. The page has the raw material to make this concrete: channel recaps, thread summaries, AI search answers are named in the footnote. Build a before/after around one of those specific use cases directly adjacent to the stat, so the number lands as a felt outcome rather than a survey result.