How it works #

From the incoming message to the finished task, every matter runs the same short path — and the human sits at the decisive point.

The path of a matter #

graph LR A["Sources: mail, scan, ticket, chat"] --> B["Feeder: reads live from the source"] B --> C["AI prepares: classify, assign, draft"] C --> D["Situational picture: visible as a tile"] D --> E["Human: confirm or correct"] E --> F["Back to the source: draft, filing, status"]

Step by step #

1 · Feeders read the sources. For every source there is a small reader that reads live what is open — without moving the data. New source, new feeder; the principle stays.

2 · The AI prepares. Every matter is classified (what is it about, how urgent), assigned (who is responsible) and, where it fits, given a draft reply. The result is a suggestion with visible reasoning.

3 · The situational picture shows it. The suggestion appears as a tile — with the most important number, a preview and the way into depth.

4 · The human decides. One click confirms. If something doesn’t fit, the draft is corrected — and Holotopos remembers the correction. Rejecting is just as fast.

5 · Back to the source. Only after confirmation does anything happen: a draft lands in the inbox (never sent automatically), a scan is filed, a status is set. The source stays the source of truth.

No autopilot. Between suggestion and effect there is always a human. That costs seconds — and keeps responsibility where it belongs.