Developers

AIEP is a protocol that makes AI reasoning verifiable by forcing models to produce evidence, reasoning steps, and dissent signals.

This is the entry point. Everything else expands from this idea.

Where AIEP sits in the AI stack

                APPLICATIONS
                     │
         Forecast / Enterprise Systems
                     │
                     ▼
                  AGENTS
                     │
                    PIEA
              (reference agent)
                     │
                     ▼
          AIEP REASONING PROTOCOL
                     │
      Evidence Rails • Reasoning Chains
      Dissent Signals • Semantic Branching
                     │
                     ▼
       EVIDENCE INFRASTRUCTURE
                     │
       Mirror Network / Evidence Index
                     │
                     ▼
          INTERNET DATA SOURCES

Three concepts

You only need to understand three things to work with AIEP.

Evidence Rails

Every answer must cite verifiable sources. Evidence is hash-bound — the protocol records not just what sources were used, but cryptographic proof that the content hasn’t changed since retrieval.

Reasoning Chains

Each reasoning step is recorded and replayable. Given the same evidence and schema version, any AIEP node produces the same canonical output. A third party can replay and verify without access to model weights.

Dissent & Branching

When uncertainty exists, alternative interpretations are surfaced — not discarded. The system records what it doesn’t know (a NegativeProofRecord) as well as what it does. Absence is proven, not merely asserted.

The 10-minute path

The fastest way to understand AIEP is to run it:

Step 1 — Clone the PIEA repository
Step 2 — Run PIEA locally
Step 3 — Ask a question
Step 4 — Observe reasoning steps and evidence rails

Once you see:

Query → Evidence retrieval → Reasoning chain → Dissent signals → Final response

The architecture becomes intuitive. Continue from PIEA Quickstart.

The ecosystem

             AIEP PROTOCOL
                   │
        ┌──────────┼──────────┐
        │          │          │
      PIEA      Mirror     Genome SDK
     Agent      Network     Tooling
        │
        │
     Applications
    (Forecast, etc.)

AIEP vs traditional AI systems

Specification architecture

The full specification set spans P01–P903+. The open-source layer consists of 137 specifications published under Apache 2.0:

Full index: SPEC_INDEX.md · Browse specifications

Documentation path

Why developers adopt AIEP

Developers adopt protocols because they can experiment within minutes. If you can run PIEA and observe reasoning traces quickly, curiosity drives further exploration.

The architecture is particularly interesting to:

• Systems engineers — building governed AI pipelines
• AI safety researchers — verifiable, auditable reasoning chains
• Infrastructure developers — distributed evidence network

These communities tend to become the earliest adopters of new protocols.

Standards, not lock-in

AIEP is designed to be adopted incrementally without proprietary lock-in. The protocol uses open web standards: JSON, SHA-256, ECDSA P-256, ISO 8601, BCP 47, Apache 2.0.

Getting started:

• Mirror Adoption — publish /.well-known/aiep/ on any site you control
• Genome SDK — schema pinning
• AIEP Node emulator — hardware trust substrate reference implementation

As adoption grows, the web evolves from a collection of pages into billions of structured artefacts that machines can retrieve and verify.

This is the point of Mirror: it makes knowledge machine discoverable at the source.

The developer role

Developers are not just building apps. They are helping build the knowledge layer of the web.

Developers: decentralise knowledge. Disseminate it.

Validate your implementation: /validator