Before you modernize a process, you need to understand how it actually works. chat2bpmn interviews the people who do the work and produces a standards-compliant BPMN diagram in a fraction of the time.
How it works
An AI agent captures how people actually do the work, in their own words, one interview at a time.
Common steps are normalized and differences are preserved as explicit findings instead of being averaged away.
Turn the reconciled workflow into exportable BPMN, process JSON, flowcharts, PRD requirements, PlantUML, or Jira-ready outputs.
Applications come in through the portal each morning. I check the passport against the file, then collect the fee, and stamp and close the case.
Cases come in from the European queue. I verify identity documents, then route to the fraud unit first — that's a step the others don't have. Then collect the fee and approve.
Renewals arrive in my queue. I confirm the documents, collect payment, and pass it along — I don't do the final approval, that goes to a supervisor.
Market landscape
Existing tools cover pieces of the problem. AI interview platforms capture conversations. Process mining tools map system logs. Requirements tools structure outputs. None of them connect all three — and none synthesize across multiple people.
|
AI Interview Platforms1 |
Process Mining Tools2 |
Requirements Tools3 |
chat2bpmn | |
|---|---|---|---|---|
| Conversational AI interview | ✓ | — | — | ✓ |
| Speech-to-text input | ✓ | — | ✓* | ✓ |
| Multi-person synthesis & divergence | — | — | — | ✓ |
| Structured workflow output & visualization | — | ✓ | ✓ | ✓ |
| Software requirements orientation | — | — | ✓ | ✓ |
1. Otter.ai, Grain, Fireflies 2. Celonis, KYP.ai, Nintex 3. Jama, Azure DevOps, Jira * Partial coverage