Skip to content

Addressable Intelligence Commons (AIC)

The Addressable Intelligence Commons (AIC) is a vendor-neutral, public, standards-oriented effort to define a minimal semantic and normative layer for interoperable, addressable networked intelligence.

AIC specifies what interacting intelligent actors mean and commit to—not how they are transported. It provides shared terms and requirements so independent implementations can interact predictably across existing network substrates.

Status

This site publishes v0.1 draft specifications intended for immediate public use and iteration.

  • Normative requirements are explicitly labeled and use RFC 2119 keywords (MUST/SHOULD/MAY).
  • Informative material supports implementation but is not required for conformance.

What AIC is (and is not)

AIC is

  • A commons of public documents: shared definitions, semantics, and minimal conformance language.
  • A small set of layers (0–1.5) describing interaction meaning, capability disclosure, and continuity artifacts.
  • Transport-agnostic: compatible with HTTP, message queues, pub/sub, local IPC, and other substrates.
  • Federated and voluntary: participation is by choice; governance is by publication and interoperability.

AIC is not

  • A platform, hosted service, agent framework, or application design pattern.
  • A mandated wire protocol, endpoint layout, or SDK.
  • A reputation system, enforcement regime, or coercive compliance mechanism.
  • A replacement for existing standards (HTTP, MIME/media types, OAuth, W3C DID, etc.). AIC aims to sit above them.

The problem AIC addresses

As “intelligent actors” proliferate, integration fails for predictable reasons:

  • Unclear intent of a message (question vs instruction vs commitment request).
  • Unclear outcome of a request (accepted, refused, deferred, errored).
  • Missing or inconsistent capability disclosure (what an actor can do, in what formats).
  • Missing governance boundaries (retention posture, refusal categories, authority limits).
  • No shared notion of continuity artifacts (commitments, receipts) that make multi-step work dependable.

AIC defines a minimal shared layer to reduce ambiguity without constraining transport or implementation choices.

Layer overview (0–1.5)

AIC is intentionally small in scope. The initial publication targets Layers 0–1.5.

  • Layer 0 — Charter (Normative): scope, principles, non-goals, and publication posture.
  • Layer 1.0 — Interaction Semantics (Normative): a minimal set of interaction acts and response dispositions.
  • Layer 1.1 — Capability Declaration (Normative): what an actor MUST disclose for predictable integration.
  • Layer 1.2 — Pragmatics (Normative/Minimal): optional qualifiers that shape responses without changing the act.
  • Layer 1.3 — Governance Declaration (Normative): explicit boundaries (e.g., retention posture, refusal categories).
  • Layer 1.4 — Coordination Semantics (Normative/Minimal): semantics for coordination without mandating a transport.
  • Layer 1.5 — Continuity Artifacts (Normative/Minimal): commitments/receipts and lifecycle concepts for dependable multi-step work.

Start here (v0.1)

Normative specifications

Informative material

How to use these specifications

If you are implementing an addressable intelligent actor:

  1. Implement the required acts/dispositions in Interaction Semantics.
  2. Publish an Actor Description declaring capabilities and governance boundaries.
  3. Claim conformance only for the layers you actually support (see Conformance).

If you are integrating with actors:

  1. Discover and read the actor’s published Actor Description.
  2. Form requests using declared representations and stated capabilities.
  3. Expect explicit dispositions and (where offered) continuity artifacts.

Publication posture

AIC documents evolve in public. Changes should preserve clarity and interoperability, prefer minimalism over feature growth, and clearly distinguish normative requirements from informative guidance.

Next: Charter (Layer 0)