What is AIEP?

AIEP — Architected Instruction & Evidence Protocol — is an open protocol for adding a verifiable evidence layer to any digital system.

SaaS platforms, databases, websites, and AI products use AIEP to publish, retrieve, and validate records that are machine-readable, hash-bound, and independently replayable. The protocol does not require AI to function — any system that produces, stores, or communicates records can adopt it.

The problem

Systems produce records but cannot prove them to third parties or machines. Outputs are trusted on faith. Evidence trails are reconstructed after the fact. AI systems have the same problem at scale, plus the prediction layer.

The core issue is not that systems are wrong. It is that there is no standard way to check:

What evidence a record was based on
Whether that evidence has changed since it was produced
Whether the reasoning can be replayed and challenged

AIEP addresses all three.

What AIEP does

Capability	What it means
Evidence binding	Every AI response is linked to the source artefacts used to generate it
Hash verification	Each artefact is hashed (SHA-256) — so any change to the source is detectable
Validation structure	A machine-readable validation result is included with every response
Replayable reasoning	The evidence rail can be re-fetched and re-validated independently
Machine-readable publication	Artefacts are published at `/.well-known/aiep/` in schema-conformant JSON

A simple example

Question:  How can an AI answer be verified?

Answer:    An AI answer can be verified by binding claims to evidence
           artefacts, hashing those artefacts, and validating the
           response against a declared evidence rail.

Evidence:
  [ev_001]  source: aiep.dev/schema/evidence.json
            hash:   f3a2c4e0b6a7d991...
            status: passed

Validation:
  Schema valid:           passed
  Hashes present:         passed
  Evidence rail present:  passed
  Status:                 PASSED

This is what an AIEP-verified response looks like. The answer is not just text — it carries its own proof.

What AIEP is not

Not a new AI model — AIEP is a protocol layer, not a model
Not a content platform — it defines how outputs are structured and verified
Not proprietary — open use under Apache 2.0, no registration required

Publish a Mirror

AIEP is two-sided. You can verify AI outputs — and you can make your own content verifiable by AI systems.

An AIEP Mirror is a machine-readable layer you publish at /.well-known/aiep/ on your domain. Five JSON files on any web server. No backend. No registration.

Once published, AI retrieval systems can discover your content directly, verify artefact hashes, and cite it with confidence — instead of scraping and guessing.

Publish a Mirror →

Try it

Verification Playground → — see a full AIEP verification flow

View a verified response example →

machine endpoint → — the live, crawlable AIEP discovery index

GitHub → — open source