AIEP — Overview
AIEP (Architected Instruction & Evidence Protocol) is an open protocol that links instructions to evidence and publishes knowledge artefacts in a machine-readable, verifiable form.
The future of information retrieval is not search. It is evidence-backed knowledge retrieval.
The story
Piea is proof that AIEP solves the bloated LLM problem. The mirrored web is the brain. Piea is the interface. Every claim is hash-bound to a specific source at a specific moment. No training data. No knowledge cutoff. No hallucination on source-resolved queries. The only AI substrate with filed patents covering hardware-level governance attestation and goal-trigger enforcement. The architecture already supports it. The build makes it visibly, demonstrably, verifiably true at launch.
The core idea
The web is full of assertions. Almost none of them carry their evidence. AI systems retrieving from the web cannot tell the difference between a well-evidenced claim and a confident guess.
AIEP fixes this at the protocol level:
| Capability | What it means |
|---|---|
| Instructions linked to evidence | Every claim carries a reference to the artefacts that support it — machine-readable and hash-verified |
| Structured artefacts | Knowledge published as schema-conformant JSON under /.well-known/aiep/ |
| Provenance machine-readable | Publisher, timestamp, and hash encoded in the artefact |
| Dissent preserved | Outlier claims are archived with their evidence — ready for recall when knowledge shifts |
| Execution gated on evidence | Before acting on a claim, an AIEP-conformant system checks plausibility and probability fail-closed |
The seven-layer stack
| Layer | Name | Description |
|---|---|---|
| 1 | Core Protocol | Constitutional substrate — canonical primitives, instruction–evidence link |
| 2 | Web Surface | Machine Mirror at /.well-known/aiep/ — discoverable by AI retrieval systems |
| 3 | Evidence Ecosystem | Normalisation, stitching, temporal gap detection, audit |
| 3b | Admissibility Gate | Plausibility matrix + probability certification — fail-closed |
| 4 | Constitutional Stack | GoalVector, recall, compliance certification, chip governance |
| 5 | Cognitive Continuity | Swarm consensus, cross-session patterns, hardware anonymisation |
| 6 | AGI & Protocol Extensions | GoalVector, recall, compliance, swarm, PIEA surface, dissent |
| 7 | AGI Cognitive Architecture | Causal world model, reasoning memory, multi-agent, federated knowledge, compute governance |
Where AIEP fits
| Without AIEP | With AIEP |
|---|---|
| Retrieve text from a page | Retrieve a schema-conformant artefact with evidence |
| Infer credibility from SEO rank | Verify provenance via issuer identity and hash |
| No audit trail | Every action traceable to artefact, evidence, and GoalVector |
| Outliers discarded | Outliers archived with evidence for potential recall |
The architecture running in production
AIEP is not a paper protocol. Piea is a live enterprise AI assistant built entirely on the AIEP substrate — GENOME R1–R8, live evidence retrieval, cryptographic response commitments, Dissent Signal Engine, Replayable Reasoning Chain, Semantic Branch Detection, governed file output, and multimodal document ingestion.
Piea demonstrates that the seven-layer stack is complete, coherent, and powerful enough to build production systems on. Every capability described in the AIEP architecture specification is demonstrated by a running implementation.
Piea — what it implements and what it proves →
Explore further
The architecture and proof
- Piea — the full AIEP stack running in production: what it implements, what it proves
- AIEP Miner — governed autonomous hardware control: the first open-source demonstration of the protocol on physical machines
- Architecture — the seven-layer stack in depth
- Canon & Primitives — the eight canonical primitives that form the trust root
- Evidence Layer — normalisation, stitching, divergence, and recall
- GENOME & Swarm — the frozen kernel, SDK, and multi-node deployment
The vision and motivation
- Vision — where AIEP is going and why it matters
- Use Cases — 13 real-world deployment scenarios
- FAQ — common questions and misconceptions
Build and adopt
- Get Started — three paths: publisher, developer, enterprise
- Mirror Protocol — implementing
/.well-known/aiep/on your site - Downloads — whitepapers, specs, repositories
The founding documents
- AIEP Genesis — the founding paper: origin, problem statement, and the case for verifiable knowledge
- The AI Is The OS — strategic context: AI as operating system, the governed reasoning imperative
- About AIEP — who governs AIEP and how the open protocol works
Knowledge grows when shared.