Control in Practice: Settings, Start to Finish

Roles and permissions are set first

Before the first document is processed, the organization defines who can see what. Intake staff can submit and view their own queues. Reviewers can open the document types their work covers. Administrators manage the rules themselves. Role-based access is enforced on every document and record in the Outcome Repository™ — not as a guideline, but as the mechanism that decides whether a file opens at all.

The policies go further than viewing: they define who can run what, on which data, in which deployment tier. Permissions are the constitution the rest of the system operates under.

The boundary is chosen per workload

Next, the organization assigns tiers. Its most obligated material — case files carrying privilege — is set to a private deployment, where the model runs on infrastructure dedicated to it and the data never leaves its control boundary. Routine correspondence runs at a lighter hybrid tier. The routing between tiers is policy-based: the platform sends each workload where its data belongs, so no employee has to make that call document by document.

Geography is part of the same decision. Deployment supports GDPR data residency requirements, and the platform can be deployed in the AWS region the organization's regulations require — residency is a setting chosen at this stage, not a question asked after the fact.

Private AI processes inside the boundary

With the settings in force, work begins. A packet arrives and MaxxOwl™ SI classifies it, extracts its data, and validates it — all inside the assigned tier. The intelligence has come to the data; the data has not gone to the intelligence. And nothing in the packet is used to train anyone's model: the organization's documents improve the organization's outcomes, a commitment held contractually, in writing.

Every access joins the record's history

As the packet moves — processed, reviewed in Outcome Workspace™, approved — each event is written into its history: what ran, on what data, when, and under whose authority. A reviewer opening the file is logged the same way the extraction was. The history belongs to the record itself, so it survives staff changes, reorganizations, and the fading of everyone's memory of a busy quarter. Months later, when an auditor asks where a document went and who touched it, the answer is read off the record rather than reconstructed from memory and inboxes.

Retention applies — and the exit works

The organization's retention rules apply on schedule: records are kept for the periods its obligations require, under the same rules in every department, because retention, classification, and access are applied consistently by the platform. And when the organization asks for everything back — for a migration, an audit, or a change of direction — a complete export of its documents, data, and histories is produced from its own repository. Control that stops at the exit would not be control.