Loading video...
Video Failed to Load
Building an Agent Team on Buzz (Complete Guide) Buzz is a new free platform, that feels exactly like Slack, except the agents aren't add ons, they are equals. and in 5 minutes you can create an agent team. And it works with your existing Codex and Claude Code subscriptions...... show more
388,759 views • 1 month ago •via X (Twitter)
33 Comments
Buzz with origin across systems and legacy messaging tools with this integration 🐝
Please follow @hot_town on x and on YouTube. Here's the youtube video if you prefer that:
Really enjoyed coming on to the pod, Riley, and it was impressive seeing how you use these tools. Keep up the great work!
Great video! inspiration for "Fizz" agent. 🐝
I’ve been messing all day with different buzz configurations excited to watch
Look, genuine question: how does this really differ from just installing something like a multi-model harness, such as T3, or even Codex OpenCodex proxies, then launching sub-agents with different models to review each other's work, provide differing opinions, challenge assumptions, etc.?
Agent team in 5 minutes is easy. Who owns the kill switch when Codex and Claude disagree on a prod write?
Good
It's as clanky as OpenClaw was in the beginning.
这平台听起来像给Agent们开了个平等会议室，5分钟组队
Treating agents as equals instead of add-ons is a real shift.
This complete guide is pure gold. Buzz finally treats agents as real teammates, not plugins — and your walkthrough makes it dead simple to build a working team in minutes. Already spinning up my first Codex + Claude channel. Huge thanks for putting this out. 🐝
The recurring tasks wall you hit at 36 mins... workflows already take cron and interval triggers 🎉 still rough around the edges but firing for me
How does the mobile app handle complex workflows?
how does buzz stop two agents changing the same files at once?
curious how codex holding the lead agent role actually changes the output compared to claude code
It’s a desktop dashboard for programs and agents @jack
At 37:23, you hit the real ceiling: scheduled agents only help if you trust they fire. Same hidden tax here. My recurring jobs live across Hermes, Codex, Claude Code, Cowork, and now Buzz. No single view of what should run, when, or where it came from. After your video, I had Fizz inventory the whole stack and mirror it into Google Calendar. One view of what was scheduled and how often. It surfaced 23 recurring series. I cut 9 redundant local jobs that were costing me tokens and noise. That same pain is why I built UltraBoard: one trust layer for recurring work across the stack.
@hot_town seen good instance of running Buzz for now in parallel as personal harness / workspace for work with own agents, then continuing with Slack for team/clients as that’s where everyone already is?
Agents-as-equals still need a human gate. Otherwise Claude Code + Codex just multiply conflicting PRs from three rooms.
Silent killer
This was a nice overview and I liked your thinking guys on how will knowledge workers work.
browse for real browsers 👀
bro why you invite people when you just want to talk all the time yourself? this is really disrespectful. let guests TALK.
The "agents as equals not add-ons" framing is doing a lot of work here. Slack's actual problem wasn't the UX, it was context fragmentation across threads. Curious whether Buzz solves that or just makes it prettier to manage 5 agents who still can't share state without you
this is cool. i have seen people talking about this but havent seen the detailed instructions like this. thank you!
where does buzz store shared context, that is where agent teams usually break
Treating AI agents as first-class team members rather than simple integrations is a massive shift in workflow design. When agents can communicate asynchronously like human devs, it dramatically reduces the cognitive load of managing complex, multi-step codebases. This looks like the logical next step for lean engineering teams.
how is it operationally any different than using hermes via slack or discord? no mention of this...
the goods! Sick man, thanks for making this video. Will be stashing for the weekend
@weeb3dev is this the video you mentioned?
Riley this is exactly right. treating agents as first-class team members is the move. I saw LangChain data, 57% of pros have agents in production now. weve been building similar agent workflows at our company, the coordination hub idea is smart.
Treating agents as equals in a Slack like surface is interesting until someone has to own the shared state and conflict resolution. That layer is still missing.
