What Taskpath is and how it works

Taskpath is permission-aware business-process automation that runs over your own open infrastructure — email, shared folders (WebDAV/Nextcloud), web forms, and webhooks — without moving your work into Microsoft, Google, or Atlassian.

The governed loop

Every request becomes a Task and moves through one governed loop from intake to done:

1. Intake — a message arrives on a channel (an email to an assistant mailbox, a web form submission, a webhook, or a file dropped in a watched folder) and becomes a Task.
2. Interpret & gather — Taskpath reads the request, pulls relevant knowledge (with citations), and asks people for anything missing.
3. Draft — it proposes an answer, a reply, or an action, always grounded in cited sources.
4. Approve — a human approves anything consequential. Nothing is sent to a third party or changed in a system of record without approval.
5. Execute & record — the approved action runs, and every step is audited.

What makes it different

• Runs on your infrastructure. Your mailbox, your files, your servers. Data stays where it already lives; Taskpath keeps only a working copy it can purge.
• Permission-aware and cited. Answers carry citations; sender identity is treated as advisory; consequential steps are human-approved and logged.
• EU-sovereign by default. Self-hostable, open infrastructure, EU-hosted AI providers — no dependency on US hyperscalers.

The three starter workflows

• Ask the organization — answer a question from indexed knowledge, cited, or escalate to a person.
• Prepare a reply — draft a grounded reply to a customer message and park it for a human to review and send.
• Process a folder — classify and act on documents dropped into a watched folder.