mine.pflow.xyz

See how your process really runs.

Give it an event log from any system that records what happened and when. Get back a map of the process as it actually ran, and a report of where the work left it.

You give it

An event log

XES or CSV. Each row says which case, which step and when.

You get back

The real process

A process map discovered from your data: the common paths, the rare ones and the loops.

You get back

Where it went off course

How many cases fit the process you expect, which ones did not, and how far off they were.

Ask from your AI assistant

This is being built as an MCP server, so Claude or another MCP client can read a log and answer questions like "which orders skipped approval?" with evidence from the log.

Status

Early. This page is all that runs here today, and nothing above works yet. The first release is laid out in the roadmap.