Where We Stand | EthosGuard
Request access Book a call
โ— Where We Stand ยท last reviewed 15 October 2026

What runs today, how far it has been tested, and what is not built yet

We ask organizations to show evidence for their AI. This page holds us to the same rule. Every capability carries two labels: how available it is, and how much testing stands behind it.

Availability

Generally availableLimited pilotDemoPlanned

Demo means it runs in the private evaluator. Limited pilot means it is offered only inside a pilot. Nothing in the product is generally available yet.

Evidence

Tested internallyTested with a clientIndependently assessedNot yet tested

Tested internally means on cases we designed ourselves. Nothing has yet been tested with a client or assessed by an independent party.

The product

What the evaluator does

Capability
Availability ยท Evidence
Scope and limits
Ten-dimension evaluation of a single AI answer
DemoTested internally
Runs in the private evaluator, open by request. A score describes one answer in its context, not the AI system that produced it.
Overall score, risk level and safety gates
DemoTested internally
Computed by code from the ten judgments. The same judgments always give the same result.
Flags with the reason behind each one
DemoTested internally
Each flag names the dimension it comes from and why it was raised.
Suggested revision of a flagged answer
DemoTested internally
Drafted for a person to approve or reject. It is not yet drawn from a client's own documents.
Citing regulatory sources in a finding
DemoTested internally
An early curated set of sources. A citation means the source was consulted, not that a regulator agrees with the finding. A fuller retrieval layer is being built.
Drawing on a client's own policies and documents
PlannedNot yet tested
Planned as part of the retrieval layer. Not available today.
Review queue with a record of each human decision
DemoTested internally
Who reviewed the case, what they decided, and when.
Audit record on every evaluation
DemoTested internally
Model, prompt version, sources consulted and time, under one record ID.
API for sending outputs from your own systems
Limited pilotNot yet tested
Offered inside a pilot. Public documentation is not published.
Exportable evidence report for an auditor or insurer
PlannedNot yet tested
A sample report shows the intended format. Export from a live account is not built yet.
Read-only API and MCP connection for your own dashboards
PlannedNot yet tested
On the roadmap. No date set.
The methodology

What is public, and what is not yet

The ten dimensions and what each one asks
Published
Evaluation Methodology v1, 15 October 2026. Read it.
How the score is computed
Published
Weights, safety gates and the audit record are described on the Methodology page.
What we are still testing
Published
Listed openly on the Methodology page, with no claimed results yet.
Clause-by-clause mapping to the EU AI Act, NIST AI RMF and ISO/IEC 42001
Not published
Drafted internally. It is published only once each line states the requirement, the evidence EthosGuard gives, and what it does not cover.
The company

Who is here, and what is still missing

Legal entity
In formation
The registered entity, jurisdiction and ownership will be stated here once formed.
Data storage and hosting
Published
Hosting, encryption, model providers and retention are on the Security and Data Handling page.
Founder
In place
Rabbi Amichai Cohen. See About.
Strategic advisor
In place
Dr. Sean Wise.
Founding engineer
Seat open
Conversations under way.
Business co-founder
Seat open
Conversations under way.
Completed client pilots
None yet
The 90-day pilot is defined and open to design partners.
EU grant
Pending
Passed the first-stage review. The outcome is not yet known.
Certification
Not offered
EthosGuard evaluates and advises. It does not certify, and it does not issue a mark.

If a line on this page is wrong, or you think we have claimed more than we can show, write to hello@ethosguardai.com with the subject line Where We Stand. We correct the page and note the change.

See one evaluation from start to finish

The ten recorded evaluations show the original answers, the findings and the suggested corrections. The sample evidence report shows the record a reviewer receives.

See ten recorded evaluations Sample evidence report Book a call
MethodologySample evidence reportPilotSecurityWhere We StandWhat EthosGuard is notAbout

Relevant frameworks: EU AI Act, NIST AI RMF, ISO/IEC 42001, OECD AI Principles. EthosGuard helps organizations evaluate AI outputs, document judgment, and keep evidence for their governance work. It does not provide medical, legal, or financial advice, does not replace qualified professional review, and does not itself establish legal or regulatory compliance.

ยฉ 2026 EthosGuard. All rights reserved. ยท hello@ethosguardai.com