Abbé Faria Partner brief / 02

A secure educational application for correctional platforms

A learning assistant the institution can govern.

Abbé Faria brings modern educational intelligence into controlled environments without introducing unrestricted internet access. Its first application is designed to work only from approved material, with deployment, policy, audit, and shutdown remaining under institutional control.

Examine the first application Review the control model
  • Approved knowledge only
  • Local or facility-edge inference
  • Operator controlled
Conceptual secure workstation with a rugged tablet, local compute appliance, network appliance, and portable power cartridge
First deployment model / 01 One bounded application · A larger system beneath it
See the first proof
01 First application

A deliberately narrow beginning

One useful thing. Completely bounded.

The first Abbé experience is designed as an educational assistant operating exclusively against an institution-approved body of material. It can explain, quiz, cite, and guide without reaching the public internet.

It is presented as one correction-ready application—not a new operating system. Underneath, the same architecture can later support a broader learning and re-entry workspace without changing the trust boundary.

Abbé Faria concept mark: a human figure before a gateway, surrounded by a red, green, and blue orbit with broken chains
Ask from approved material Answer with evidence Refuse beyond scope

The first product promise

  • 01
    Bounded knowledge

    The assistant responds from an approved corpus and clearly identifies the material supporting an answer.

  • 02
    Visible limits

    Questions outside the permitted scope produce an explicit refusal rather than an improvised answer.

  • 03
    Institutional control

    Content, access, logging, updates, reset, and disable behavior are defined with the operator.

  • 04
    Measurable learning

    One curriculum and one objective make comprehension, progression, engagement, and system behavior observable.

“The refusal is not a limitation around the product. It is part of the product.”

— Control principle
02 Control model

Nothing important is assumed

Three approvals. One controlled path.

Approved curriculum does not automatically authorize a third-party application, and an approved application does not automatically authorize local inference. Abbé treats each decision as a separate technical and institutional gate.

Concept visualization of approved signals passing through a controlled gateway into a local intelligence core and managed endpoints
No step is presumed Approved content → allowed application → permitted inference → pilot
  1. 01

    Content

    The curriculum, sources, learning objective, and prohibited material are reviewed.

  2. 02

    Application

    The runtime, package, storage, identity, updates, and device permissions are approved.

  3. 03

    Inference

    Processing is placed on-device or at a sealed facility edge, subject to explicit policy.

  4. 04

    Pilot

    A defined cohort begins only after controls, measures, failure conditions, and shutdown are agreed.

A durable platform boundary

Each party keeps the authority it should have.

  • Platform partner: distribution, device policy, identity, and facility access01
  • Facility: availability, approved content, oversight, retention, and disable authority02
  • Abbé Faria: application, models, curriculum workspace, and portable user-state layer03
  • Adapters: provider-specific integration without rebuilding the Abbé experience04
03 Expansion

A wedge, not a ceiling

The tutor proves the boundary. Then the world gets larger.

Abbé begins with the easiest capability to evaluate and govern. Each later experience inherits the same approved identity, content, policy, audit, and deployment model.

A / First proof

Approved-corpus tutor

Explanations, cited answers, practice, and progression inside one measurable curriculum.

B / Workspace

Library and creation tools

Curated knowledge joined to bounded writing, planning, design, and project work.

C / Re-entry

Practical digital workflows

Guided preparation for documents, services, employment, education, and life beyond release.

D / Second endpoint

Voice, simulation, and VR

An embodied extension of the already-governed environment—not a separate trust experiment.

04 Partner path

Fit before fleet

First answer the platform questions. Then earn the pilot.

The initial request is a short technical-fit conversation—not a fleet-wide commitment. A pilot follows only when the device generation, execution model, operating controls, and learning objective are clear.

  1. 01Identify

    Confirm the supported devices, Android versions, runtime, and application pathway.

  2. 02Place

    Decide whether inference belongs on-device or on a sealed facility-edge appliance.

  3. 03Verify

    Demonstrate source boundaries, refusal, audit, signed updates, reset, and shutdown.

  4. 04Pilot

    Test one curriculum with one facility context, defined cohort, and shared measures.

  5. 05Expand

    Add capabilities or endpoints only when the evidence and operating relationship support them.

The first partner conversation

Which current device generations and runtimes support a third-party application?

May approved inference run on-device, at the facility edge, or through another controlled topology?

How are identity, content, updates, logs, retention, and emergency disablement governed?

The larger thesis

The tutor proves the boundary.
The boundary makes the world possible.

One bounded application first. A portable, governed workspace across tablets, facility-edge systems, and future immersive endpoints next.

Discuss platform fit