Bearing · how it works

Proprietary instrument

BEARING

Ready is not the same as able to prove it.

Every AI use raises the same question eventually

A supervisor, an auditor or a client asks who's accountable for a specific AI decision. The honest answer is usually "several people, informally", which in practice means no one.

An agent, not a questionnaire

What it reads, not what people remember

The usual way

Asks whether you have the policy, and takes yes for an answer.

The Bearing agent

Finds the document that evidences each obligation, and the person who owns it.

Step one

What it connects to, read-only

Document storesPolicy libraryModel registerCommittee recordsRoles and accountabilities
Bearing agent

Read access only. Nothing is changed in your systems.

Step two

How it investigates

connectdocument stores, policy library and model register: read-only
map212 obligations across SM&CR, Consumer Duty, FINMA and the EU AI Act
find131 have a document. 81 have none. 24 are out of date.
owncredit-decisioning model has 3 named owners. Accountability is split.
verify6 of 10 strongest yes answers could be backed by a document
keepsupervisor pack updated. 2 obligations changed this month.

Every line is something a person can check afterwards.

Step three

What it produces

ObligationEvidenceOwner
Model risk management covering AIFoundNamed
AI inventoryMissingNone
Consumer Duty outcome testingOut of dateNamed
Human oversight of AI decisionsFoundSplit

An obligation-to-evidence map.

How it is held to account

Rules, not vibes

01

A yes is not evidence

An answer counts only when the document behind it can be produced.

02

One accountable person

Split or missing ownership is reported as a finding, not smoothed over.

03

Changes are tracked

When the regulation or the evidence changes, the supervisor pack changes.

The agent is held to the same questions it asks of everyone else.

Configured to your regulator

The same agent. Your market.

SM&CR and Consumer Duty obligations, with PRA model risk expectations.
FINMA's guidance and the Federal Act on Data Protection.
EU AI Act obligations by risk class, with DORA where it applies.

Choose a market. Other markets are added as packs.

How an engagement runs

From first conversation to a working agent

1

Scope

Which regimes, which entities, who owns what.

2

Connect

Read-only connections to your document stores, reviewed first.

3

First evidence map

Every obligation, the evidence found and the owner named.

4

Keep the pack current

Stale evidence and regulatory change are flagged.

A typical shape. The sequence and timings are set with you.

One platform

Bearing does not work alone

KeelSets the policy and the gates for every use case.
SentinelQuestions what each agent can see, change and cause.
SoundingsSees what is actually running across the organisation.
BearingProves it to the supervisor, obligation by obligation.
The human layerHALO, Squall and Ensign measure what no system log can.

See the whole system on the Solutions page.

That's Bearing

Find out where your firm actually stands.

A demonstration of the method is one click away. The deployed Bearing is built for your environment and configured to your regulator.

Discuss a deployment