Meridian Become a design partner

PUBLIC CLAIM LEDGER / 15 August 2026

Meridian does not grade Meridian.

Every material claim belongs to an evidence class. Internal engineering results can justify another test. They cannot create external authority, customers, or economic impact.

Inspect current stateTry to break Meridian
Evidence classes never collapse into one score.INTERNAL != EXTERNAL != COMMERCIAL != ECONOMIC

CURRENT EVIDENCE STATE

What Meridian may claim today.

The ledger advances only when qualifying evidence exists. Zero is not hidden, softened, or converted into a synthetic proxy.

01 / INTERNALDEMONSTRATED
8

Controlled evaluation suites passed

Useful engineering evidence produced inside Meridian's own controlled environment.

May support: capability hypothesis
02 / EXTERNALUNPROVEN
0

Independent intelligence suites passed

Requires a sealed evaluation Meridian did not author, operate, or grade.

Cannot claim: independent validity
03 / COMMERCIALUNPROVEN
0

Paying organizations

Requires an authorized paid pilot or recurring commercial conversion.

Cannot claim: market validation
04 / ECONOMICUNPROVEN
GBP 0

Externally verified impact

Requires an authorized baseline, observed outcome, and external verification.

Cannot claim: value created

Last evidence refresh: 15 August 2026. Public values remain unchanged until their proof obligation closes.

CONTROLLED EVALUATION

Five hidden developments. Two thousand fragmented signals.

Across five fictional organizations, Meridian surfaced every planted development before explicit confirmation. The system preserved the evidence chain, consequence, intervention, uncertainty, and authority boundary.

SCOPE BOUNDARY

Synthetic evidence. Fictional organizations. Controlled test. No customer outcome or production performance is implied.

CASEDEVELOPMENTEARLY BYEXPOSURE
NB-01Client renewal riskNorthbridge Studio116 signalsGBP 180K
RW-02Margin leakageRedwood Advisory129 signalsGBP 120K
HM-03Delivery concentrationHarbour Managed Services115 signalsGBP 95K
EW-04Decision bottleneckElmworks Engineering142 signalsGBP 80K
LS-05Revenue opportunityLumen Software118 signalsGBP 250K
RESULT5 / 5 surfaced115-142 signals earlyGBP 825K synthetic

FAILURE LINEAGE

A perfect score triggered suspicion, not celebration.

Meridian preserves adverse results. An early blind suite scored 100%. Harder tests then exposed weaknesses, and the public lineage shows exactly what broke.

01
Blind suite

Precision 100% / Recall 100%

PASS
02
Adversarial attack

Precision 78.6% / Recall 73.3%

FAIL
03
Semantic frontier

Generalization weakness exposed

FAIL
04
Role-binding holdout

Successor weakness repaired

FAIL
05
Temporal holdout

25 precommitted scenarios passed

PASS

This lineage strengthens the engineering record. It does not substitute for an independent evaluation.

BUILT SYSTEM ACTIVITY

Scale is reported as scale, not traction.

These figures describe Meridian's runtime and engineering breadth. They do not describe deployments, organizations, revenue, or economic outcomes.

614
HTTP routes
259,811
Audit records
57,120
Job runs
33,244
Permission checks

THE NEXT PROOF OBLIGATION

Do not trust Meridian.
Try to break it.

Bring one difficult operational question and a bounded evidence set. Meridian must expose its support, uncertainty, authority boundary, and accept a pass, mixed, or failed result.

Challenge MeridianBecome a design partner
No customer data is requested in the first conversation. No result receives public credit without explicit evidence and authorization.