v0.9.0: Automation Workflows โ€” let Paca hand off tasks for you ยท Paca-AI ยท Discussion #268

ยท GitHub

2 min read Original article โ†—

๐Ÿš€ Automation Workflows are here

Paca can now automatically re-assign tasks as they move through your process โ€” no more manually pinging the next person (or agent) in line every time a task is ready to move on.

Paca v0.9.0 โ€” Automation Workflows

What it does

Build a workflow on a visual canvas: pick the tasks you want to automate, wire them together with edges, and define two simple rules:

  • Status rules โ€” "when a task in this workflow reaches status X, assign it to member M"
  • Status transitions โ€” "when work at status X is done, it moves to status Y"

From there, Paca's automation engine takes over:

  1. On every status change โ€” it looks up the rule for the new status and reassigns the task if needed.
  2. On every "done" โ€” it walks the task's outgoing edges and hands the next task off to its assignee, automatically. If a task depends on more than one predecessor, it waits until all of them are done before releasing it (AND-join).

Workflows are drafts until you activate them, so you can build and tweak the graph safely before it goes live โ€” and you can always pause (revert to draft) or archive one later.

Built for human + AI teams

Since Paca agents are first-class teammates, they can be assigned by these workflows just like a human. When an automation hands a task to an agent, it tells the agent what workflow assigned it and what status to set next โ€” so multi-step pipelines flow correctly one hop at a time instead of agents guessing the final status.

AI agents can also build and edit workflows themselves through the MCP server โ€” create a workflow, lay out nodes, wire up rules and transitions, all through natural conversation.

Get started

As always, we'd love to hear what you build with it โ€” drop questions or feedback below. ๐Ÿพ