Launching today
agentpulse
hire AI employees and run them like a real company
13 followers
hire AI employees and run them like a real company
13 followers
AgentPulse is an open-source AI company workspace. Hire AI employees, assign goals in a group chat, let them discuss before they act, and approve every high-risk move. No prompts, no workflow graphs — if you can run a group chat, you can run an AI company.
The group chat setup makes delegating to AI agents feel way more natural than dragging nodes around in a workflow builder. Curious how approval thresholds handle edge cases like data access requests.
@fahri1539449 Good question — right now approval requests surface with full context (what the agent wants to do, why) before anything executes, so edge cases like data access just become another thing you sign off on rather than something the system has to guess about. Still refining the UX around it based on feedback like yours.
If you're curious to dig deeper, the repo's open: github.com/Clycheng/agentpulse — a star would mean a lot, still early days on this 🙏
The group chat metaphor actually clicks here, and watching the AI employees debate before acting feels less like a gimmick and more like a real team dynamic. Approval gates for risky moves is a smart touch.
@sla919522062231 Really glad that landed — that "team debate before acting" feel is exactly what I was going for, it's easy to make multi-agent stuff feel mechanical instead of like actual collaboration. Appreciate you noticing it.
If you want to poke around under the hood, it's open source: github.com/Clycheng/agentpulse — stars genuinely help a solo dev keep momentum, so thank you if you drop one 🙏
A nice Slack-style approach that lowers the bar for orchestration. One thing that would really help: a shared audit log or replay view of every agent conversation thread, so you can scrub back through who said what before that big approve click. Even better if it diffs between runs so you can see why an outcome changed.
@anlpiyr This is great feedback, and honestly one of the more requested things so far — a run/activity log so you can trace back what each agent did instead of just seeing the end result. Taking it seriously as a next step. The diff-across-runs idea is sharp too, further out but noted.
Repo's here if you want to follow along or poke at the code: github.com/Clycheng/agentpulse — would really appreciate a star if it's useful to you 🙏
The group chat flow is a smart way to make this feel approachable. One thing I'd love to see is a shared activity log or timeline view that shows what each agent did in chronological order, so I can trace decisions back to the exact message that triggered them when reviewing past runs.
@zilantekatusgw Same theme a few others raised too, so it's clearly a real gap — a timeline view tracing each agent's actions back to the message that triggered them is next up on my list. Thanks for putting it into words this clearly.
If you want to follow progress, it's open source: github.com/Clycheng/agentpulse — a star would really help, this is a one-person project so every bit of support counts 🙏
Tried it out and the group chat feel really does click, kind of surreal watching the agents debate a task before settling on an answer. Took some getting used to approving high-risk steps, but that's the part I appreciate most.
@abdulkadir97114 That "took getting used to" reaction is actually the best signal I could ask for — the friction is on purpose (it's a deliberate "boss signs off" model instead of full autonomy), so hearing it clicked for you after using it means a lot.
Thanks for actually trying it out. If you're up for it, a star on the repo would help a ton: github.com/Clycheng/agentpulse 🙏