Skip to content

Ai under your control

Off until you turn it on. Labeled when you do.

Ai proposes, cites and drafts. People accept, decide and sign. It never rates credibility, recommends discipline or scores officers.

The cited timeline of a case: an Ai proposal with a dashed border offers a proposed entry with its source recording and the buttons Reject, Edit and Accept, above an entry the investigator wrote, each citing the body-camera recording and the moment in it.
An Ai proposal in the demo agency: marked, cited to its source, and waiting for a person to accept, edit or reject it. Every person and event is fictional.
  1. 01Off by default

    Every part of CoordinatedIA works without Ai. An agency administrator turns Ai on after reading a plain notice of what is sent to which provider, and can turn it off again.

  2. 02Always labeled

    Ai-generated text is marked wherever it appears, shows the sources it used, and stays marked until a person has reviewed and accepted it. Printed reports carry a disclosure while any Ai text remains.

  3. 03People decide

    Ai output is a proposal. Nothing is applied to the record, sent to anyone or signed until a person accepts it. The investigator certifies that they reviewed the content and take responsibility for it.

  4. 04Hard limits

    Ai does not rate credibility, recommend findings or discipline, identify people from faces or voices, or rank officers. Spending is capped for each agency, and the agency sees its usage.

Access is by invitation while CoordinatedIA is in early access.