ARA · Thesis

Our Canonical Manifesto · v1.0

The next layer of computing is not another model. It is the substrate beneath every model.

For half a century, software has learned everything a person does and almost nothing about how they are. That asymmetry ends here.

A public statement of intent, not an invention disclosure.

IPremise

Every era of computing added a missing layer.

Networking gave machines a way to reach one another. Databases gave them a way to remember. The cloud gave them a place to live. Mobile gave them a body. Large language models gave them a voice.

None of these layers taught a machine the one thing every human counterpart carries into every interaction: internal state. How a person is, in this moment, in a form the machine can actually receive.

This is the missing layer. Not sentiment. Not affect detection. Not another wrapper on a model. A substrate.

Technology has always understood what we do. It has never understood how we are.

IIThe Layer

The Emotional Signal Infrastructure Layer.

ESIL is a consent-scoped substrate that lets any system respond to declared human internal state as first-class context. It sits below applications, below models, below agents. It composes with all of them.

It is not surveillance rebranded. It is not affect inference. It does not read faces or listen for stress. It receives what a person chooses to declare, in a pre-verbal form designed for the substrate, and makes that declaration machine legible while respecting personal processing boundaries.

A consumer surface generates the corpus. An SDK makes the corpus useful. Neither works without the other. Together they are one nervous system.

The public experience and the infrastructure share a commitment to voluntary expression. Proprietary mechanisms remain private.

IIIPrinciples

What we will not compromise.

  1. 01

    Declared, never inferred.

    No facial analysis. No voice stress. No biometric scraping. Signal enters this substrate only when a person chooses to place it there.

  2. 02

    Attested, not identified.

    Trust in an emotional system belongs to the continuity of behaviour, not to a passport. We build attestation as a trust primitive, not identity as a gate.

  3. 03

    Non-diagnostic by construction.

    The system does not label a person, score a person, or pathologise a person. It reflects. Diagnosis is a category error at this layer.

  4. 04

    Voluntary, revocable, portable.

    The person who declared the signal owns it. They can withdraw it. They can carry it. Consent is a live property, not a checkbox.

  5. 05

    Infrastructure before application.

    We build the substrate first. Applications composed on top are downstream consequences, not the point.

  6. 06

    Durable over fast.

    Every schema and every primitive is designed to still be correct in ten years. Speed that compromises this is refused.

IVProof

The substrate is already breathing.

A thesis without a corpus is a slide deck. Ours is running. The counters below show verified platform activity, refreshed periodically.

Substrate, in real time

Completed Pulse signals

Lifetime RAE outputs

Platform activity is not a claim of scientific validation.

VHorizon

What this layer becomes.

When an AI agent reads emotional context before answering, that is this layer. When a robot pauses because the person near it is not ready, that is this layer. When a piece of software treats a human as more than a click, that is this layer.

We are not building a product category. We are building the primitive every product category will eventually depend on. The companies that see this early are the ones we will build with.

The next decade of intelligent systems will not be defined by what they know, but by what they can receive.

VISignature

This document is a public record of intent. It will be revised as the substrate matures. It will not be softened.

Signed

Frank Ande

Co-founder, Chief Executive

Countersigned

Pelumi Ande

Co-founder, Head of Product Architecture

ARA · Our Canonical ManifestoVersion 1.1 · October 2026
How the layer worksWorking papersBuild on the substrate