Evidence

Inspect the retrieval, not just the claim.

Reviewed traces show how a request becomes bounded, ranked, source-linked context. Every published example is fixed to a corpus version and an evaluation record.

Published artifacts
2 reviewed
Pinned corpus
2026-08-21.13
Evaluated
2026-08-211
Endpoint
Not public

Claims and evidence

Four claims, each bounded by its evidence.

This is one evaluated EU AI Act workload, not a universal quality claim.

Every figure comes from one frozen evaluation set. It measures that set, not another corpus, language, or workload.

  1. Claim

    Retrieval finds the required evidence on this evaluation set.

    Evidence

    96.2%1Recall@10

    Dataset
    EU AI Act v1, 80 fixtures, 1,160 rows
    Base
    70 answerable queries
    Date
    2026-08-21, private-beta endpoint
    Method
    How much of the required evidence appeared in the first ten retrieved passages.
    Limit

    Finding is not ranking.

    Ordering is measured separately: nDCG@10 82.0%, MRR@10 78.9%.

  2. Claim

    Results point to the official article or annex anchor.

    Evidence

    98.3%1Source-anchor recall@10

    Dataset
    EU AI Act v1, 80 fixtures, 1,160 rows
    Base
    70 answerable queries
    Date
    2026-08-21, private-beta endpoint
    Method
    Whether the expected official article or annex anchor appeared in the result set.
    Limit

    Focus is lower than recall.

    Citation precision@10 is 53.4%. It measures focus on required anchors, not legal correctness.

  3. Claim

    Unsupported queries return an explicit abstention.

    Evidence

    100%2Abstention accuracy

    Dataset
    EU AI Act v1, 80 fixtures, 1,160 rows
    Base
    10 unsupported queries
    Date
    2026-08-21, private-beta endpoint
    Method
    Whether unsupported queries returned an explicit abstention instead of unrelated context.
    Limit

    Small denominator.

    Covers the 10 unsupported fixtures in this set only. It does not establish abstention on another corpus or workload.

  4. Claim

    Every returned passage stayed inside the enforced scope.

    Evidence

    01Boundary violations

    Dataset
    EU AI Act v1, 80 fixtures, 1,160 rows
    Base
    80 recorded responses
    Date
    2026-08-21, private-beta endpoint
    Method
    Returned passages outside the enforced source scope. Tracked separately from quality scores.
    Limit

    Frozen corpus, private endpoint.

    Applies to this frozen v1 corpus. The retrieval endpoint is not publicly available.

One published query

How a question becomes cited evidence.

CorpusMesh returns retrieval context for a customer-controlled model. Each result keeps its citation label, EUR-Lex anchor, corpus version, verification scope, and lineage evidence. The service does not generate the final answer.

Caller-controlled

How do Article 6 and Annex III work together to identify high-risk AI systems?

The question enters as text, not as a client-made vector.

Knowledge base
eu-ai-act-reference
Requested filters
language en · jurisdiction eu
Result limit
top_k 10
Profile
gemini-embedding-2-768-search-v2

Published source trace, reviewed 2026-08-213

Inspect the cited EU AI Act trace

Publication register

What is published, and what is held back.

One row per artifact. The internal version stays listed so the boundary of what is public remains visible.

  1. 80 EU AI Act queries, measured end to end

    Review recall, ranking, citation precision, abstention, configuration selection, and sealed reproducibility evidence.

    Inspect the RAG evaluation
    Artifact
    Published v1 RAG evaluation
    Corpus
    v1 · 2026-08-21.1
    Evaluation
    80 fixed queries · Recall@10 96.2% · 0 boundary violations
    Recorded
    2026-08-21
    Published
  2. EU AI Act: Article 6 and Annex III

    Follow one measured request through mandatory scope, five ranked passages, official EUR-Lex citations, and sealed lineage.

    Inspect the cited EU AI Act trace
    Artifact
    Published v1 · RAG with citations
    Corpus
    v1 · 2026-08-21.1
    Evaluation
    80 fixtures · 0 boundary violations
    Reviewed
    2026-08-21
    Published
  3. Version 2026-08-24.3 trace

    Active internally. Its trace is not authorized for public publication.

    Artifact
    Internal version trace
    Corpus
    2026-08-24.3
    Access
    Internal only
    Not published
Publication boundary

These reviewed public artifacts remain pinned to v1. Version 2026-08-24.3 is active internally, but its trace is not authorized for public publication and the retrieval endpoint is not publicly available.

What this does not prove

Retrieval evidence is not a legal answer.

  • The retrieval endpoint remains restricted to an internal bearer token.
  • The trace returns source context and does not generate or validate a legal answer.
  • The internal reference corpus is not an available commercial knowledge base.

Where each figure comes from.

  1. Runtime endpoint report

    Recorded 2026-08-21 on the deployed private-beta endpoint, 80 responses.

    Deployed commit 57c28e321dbddcad4eb57006d4015a5c1e0e204f

    Report SHA-256 052cbcd7621f56e6d37b7addd7c931234af2d0f1422948bc3c29210eb2500c7f

    Inspect the RAG evaluation
  2. Sealed fixture set and corpus

    80 fixed queries on 1,160 document rows, cutoff 10. Split 50 calibration / 30 validation.

    Fixture set SHA-256 c1780012a679b35b8d43e32bd8be4149af391be7cd971a66e54d2f348fa65b73

    Chunk manifest SHA-256 2d73f00da7bd0cfbf942c26e96dbe27bbcf9161f428dd7985d6d669581e74924

  3. Published source trace

    Reviewed 2026-08-21 by CorpusMesh product and source review. Corpus version 2026-08-21.1, 5 of 10 results shown.

    Public trace SHA-256 88821ad972ee9c05ed2d56e6f125baa0000c9f9abc4515f34dd82db3ef61fc17

    Inspect the cited EU AI Act trace
  4. Official source

    EUR-Lex, Regulation (EU) 2024/1689, verified 2026-08-21.

    Open the regulation on EUR-Lex (opens in a new tab)

These identifiers tie the claims to one evaluated system state. They support provenance; this page does not provide the complete artifacts needed for an independent replay.

Build with bounded knowledge

Need maintained knowledge for your own application?

Tell us which sources, jurisdictions, and update obligations your product depends on. We will use the request to prioritize coverage, not pretend that a planned corpus already exists.

New to retrieval scoring? See how recall@k, MRR and nDCG are calculated.

Access
Private beta. Access is reviewed and approved manually.