Slack

Our verdict on Slack

We don't use Slack as an AI tool. We use it as the place our AI tools report back, because that is where our teams already are. The best example is Sam's daily check-in, a two-hour build that turned eight to ten stand-up forms into one AI summary in Slack each morning.

This page covers how we wire AI into Slack, not a review of Slack's own AI features, which we have not tested on the show.

What we built with Slack

• Daily Check-in (August 2025): Sam replaced a Google Form stand-up with his own web app. Each team says what it finished, today's priorities and any blockers. An AI summary, blockers first, posts to a founders' Slack channel a little after 10am Central, and each team's individual check-in still posts to its own channel. He built it in about two hours across a few days, launched on a Monday and shipped a v2 from team feedback that week.
• Bito Slack bot prototype (December 2025): Jordan built a Slack look-alike in Google AI Studio with a bot for Bito's code reviewer. "Bito last" pulled up a review of his last pull request, which had failed on a critical security bug. It also told jokes and played disco music. Jordan said it was "definitely not ready" for production.

How we use Slack with AI

1. Replicate the form people already fill out. Sam vibe coded a copy of the existing stand-up form in Bolt instead of using Google Forms or Typeform.
2. Store every submission in a database. He used Supabase for both the data and the edge functions.
3. Summarize with a prompt that ranks what matters. His prompt puts blockers first, because those are what a founder needs to act on before the day starts.
4. Post on a schedule. A GitHub cron job sends the summary to Slack at a fixed time every morning.
5. Keep the detail one click away. The raw check-ins still go to each team's channel, so you can dig in when the summary flags something.
6. Add metrics and history later. Sam added check-ins by date, on-time and completion rates per team, and team management.
7. Use Slack for alerts too. Jordan told Bito's Amar Goel we use Slack "all the time for notifications" and callouts when things go wrong.

So I said, why can't we just have AI summarize this and post to a Slack channel?

— Sam Nadler, Episode 7

It paid off fast. The day before we recorded, the summary showed our sales team had run out of credits for a tool, and Sam topped them up immediately instead of reading that form later in the day (Episode 7, at 10:30).

And, you know, now I'm learning how to build edge functions and cron jobs, and it's all pretty easy with AI. Cumulatively, probably took me two hours total.

— Sam Nadler, Episode 7

Where Slack falls short

• Permissions. Sam's check-in had no permissions at launch, so he asked teams to hold back sensitive information until he added them.
• Summaries don't check follow-through. Jordan's suggestion was to have the AI look back over the last seven days and see whether what teams prioritized actually got done. We have not built that yet.
• Bots need real integration work. The Bito bot was a mock-up. Amar said customers do ask for Slack notifications when a code review finishes, but that takes the real product, not a prototype.

Slack in our guests' products

Several guests treat Slack the same way. Attention's Anis Bennaceur said his product reads Slack messages as an input and delivers insights in Slack, email or the CRM (Episode 45). David Petrou put his AI agent into the places people already talk, Slack included, because people work in groups (Episode 51).

Inside our own company, Sam said in Episode 39 that teams building their own tools now integrate them with what we already use, such as Jira, Granola and Slack. Sam also dictates longer Slack messages with Wispr Flow.

Episodes featuring Slack

• Episode 7: the daily check-in and its AI summary in Slack.
• Episode 23: Jordan's Bito Slack bot, built live for Amar Goel.
• Episode 39: how teams at Ryz Labs connect their builds to Slack, Jira and Granola.

FAQ

How do you post an AI summary to Slack?

Sam stored stand-up forms in Supabase, used a Supabase edge function to summarize them with AI, and ran a GitHub cron job that posts the summary to a Slack channel each morning. It took him about two hours in total.

Can a non-developer build a Slack integration with AI?

Yes. Sam knew a little SQL before AI and built the daily check-in, including edge functions and a cron job, with Bolt in between meetings over a few days.

What should an AI Slack summary show first?

Blockers. Sam's prompt tells the AI to put blockers at the top, and the day before we recorded, it surfaced that the sales team had run out of credits on a tool so he could top them up right away.

Is AI in Slack only useful for summaries?

No. Jordan uses Slack for notifications and callouts when things go wrong, and mocked up a Slack bot for Bito that reviews your last pull request.