Epismo
Share and improve how teams work with AI
370 followers
Share and improve how teams work with AI
370 followers
In most companies, the best ways of working with AI agents stay buried in a few people’s chats and heads. The rest of the company cannot reuse them, build on them, or improve them together.
Epismo turns those methods into reusable playbooks for real work. Each playbook coordinates the people and agents involved, adds human review where needed, and records what happened so the team can improve the next run.
This is the 6th launch from Epismo. View more
Epismo Playbooks
Launching today
Your best ways of working with AI agents shouldn’t disappear into private chats. Skills alone don’t capture how real work gets done. Epismo let you share the whole method: instructions, agent harnesses like skills and graphs, context, intermediate outputs, expected deliverables, and human reviews. Hand work to another person or agent with what they need to continue — using Claude Code, Codex, OpenClaw, or whatever comes next.
Free
Launch Team
Epismo
Hey Product Hunt 👋 I’m Hiroki, the solo founder of Epismo.
This weekend, I had an idea that made me pivot the whole product.
We’ve gotten much better at sharing prompts, skills, and agent tools. But a skill alone rarely captures how a real job gets done.
Real work has a method around it: what steps to follow, which skills or graphs to use, what context and intermediate outputs to carry forward, what the final output should look like, and when a human or another agent needs to take over.
So I rebuilt Epismo around Playbooks.
A playbook packages that whole method while leaving execution to the tools you already use — Claude Code, Codex, OpenClaw, or whatever comes next.
When work moves between people or agents, the method and agent setup can move with it.
The idea clicked on Friday. I spent the weekend rebuilding the core product and deployed it right after.
So Epismo Playbooks is very new, and the library is still mostly empty.
I’d love to change that with everyone here.
There’s a lot of interesting work happening around agent harnesses, graph engineering, and loop engineering right now. I think the missing layer is a way to share the whole way of working those pieces enable — and learn from what happens when people and agents actually use it.
What’s an AI workflow you’ve figured out that other people should steal?
Share it below. Let’s build the first playbooks together.