Distributed Inference

In development · Seeking pilot organisations

Policy-controlled AI infrastructure for sensitive organisations

AI where it works.
Data where it belongs.

Use powerful AI while keeping control over where your information goes.

Distributed Inference places a policy-controlled boundary between your applications, private knowledge and AI. Applications request a capability. Policy determines where it may run and what information may leave your environment.

The file stays. Only what must leaves.

Discuss a pilot See how it works

DI / PRINCIPLE / 01

Your applications shouldn't decide where sensitive data goes.

Direct AI integrations can distribute provider credentials and leave individual applications responsible for deciding where organisational information is sent. Distributed Inference creates one controlled boundary for AI.

01

Capability, not provider

Applications describe what they need.

02

Policy, not hardcoding

Organisational rules determine where inference may happen.

03

Minimum necessary disclosure

Only context permitted by policy crosses the boundary.

DI / FLOW / 02

From request
to result.

Applications stay focused on the capability they need. Distributed Inference applies the organisation's rules around placement and disclosure.

  1. 01 / REQUEST

    Ask for a capability

    Noah or another application asks for an AI capability without selecting an external provider or holding its credentials.

    CAPABILITY  legal.reasoning
  2. 02 / POLICY

    Apply organisational policy

    Distributed Inference determines whether the request may run locally or externally, where it may be placed, and what information may accompany it.

    CLASS  LEGAL.MATTERREGION  UKDECISION  ALLOW
  3. 03 / MINIMISE

    Send only what is permitted

    Private source material stays local. Only the context required and permitted for the task is eligible to cross the boundary.

  4. 04 / PLACE

    Run AI in the right place

    The same application can use different inference locations as organisational policy requires.

    Local AIApproved frontier AIRegion-specific AI
  5. 05 / RETURN

    Bring the result back through the boundary

    The response returns through Distributed Inference to the requesting application.

  6. 06 / AUDIT

    Know what happened and why

    Policy decisions and permitted disclosures are recorded so organisations can understand how AI was used.

DI / ARCHITECTURE / 03

One controlled boundary for AI

A simple capability contract gives applications governed access to AI while organisational policy controls where work may run and what information may accompany it.

01Applications ask for capabilities, not providers.

02Provider credentials remain behind the Distributed Inference boundary.

ILLUSTRATIVE / LEGAL / 04

Private knowledge.
Frontier reasoning.

“Have we handled a matter similar to this before?”

Distributed Inference can search private organisational knowledge locally and use approved AI reasoning without uploading the firm's document repository. Organisational policy controls what context may leave and where it may be processed.

Illustrative product direction. Grounded answers and citations are planned capabilities, not a statement of current production availability.

PRIVATE KNOWLEDGE500,000private documentsIllustrative scenario
SEARCH LOCALLY
LOCAL RESULTRelevantevidence
APPLY POLICY
CONTROLLED DISCLOSUREMinimumpermitted context
APPROVED REASONING
RETURNED RESULTGroundedanswer

DI / APPLICATIONS / 05

Built for sensitive environments

Explore where policy-controlled access to AI could create value when information must remain within approved environments, regions or supply chains.

01 / PROFESSIONAL SERVICES

Legal & advisory

Search matters, engagements, contracts and correspondence while keeping client knowledge under organisational control.

  • Find similar matters
  • Review contracts
  • Due-diligence support
  • Retrieve supporting evidence
02 / HEALTH & LIFE SCIENCES

Healthcare & life sciences

Explore advanced AI while enforcing policies around clinical, patient, trial and research information.

  • Clinical knowledge search
  • Approved-document summarisation
  • Research knowledge
  • Policy-controlled assistance
03 / FINANCIAL SERVICES

Finance & insurance

Give teams governed access to AI across customer, transactional, risk and policy information.

  • Policy and procedure search
  • Case summarisation
  • Risk knowledge
  • Audit support
04 / CONTROLLED INFORMATION

Defence & public sector

Apply organisational handling rules when exploring AI over sensitive operational and government information.

  • Controlled knowledge search
  • Document assistance
  • Policy retrieval
  • Approved-environment workflows
05 / CRITICAL INFRASTRUCTURE

Energy & utilities

Keep operational and infrastructure knowledge within approved boundaries while enabling controlled AI assistance.

  • Technical knowledge
  • Maintenance information
  • Procedure search
  • Operational support
06 / INTELLECTUAL PROPERTY

Engineering & research

Work with designs, technical knowledge and commercially sensitive research without treating external disclosure as the default.

  • Engineering knowledge
  • Research discovery
  • Technical summarisation
  • Evidence retrieval

Illustrative areas for pilot exploration. Requirements, deployment constraints and permitted uses would be assessed with each organisation.

DI / FIRST-PARTY / 06

Meet Noah

The first-party AI workspace built on Distributed Inference.

Noah brings chat, search, skills, agents and workflows to private organisational knowledge, while Distributed Inference provides controlled AI capability.

Already have your own application?You don't need Noah. Existing software, plugins and enterprise systems can integrate through the API.
NOAH CONCEPTCONTROLLED

DI / API / 07

Built for more than one application

Noah is a first-party client. Third-party applications and enterprise systems can use the same capability contract—without needing to know which provider ultimately executes the request.

DI / THESIS / 08

Why we're building it

AI capability is advancing rapidly, but organisations should not have to choose between using powerful models and maintaining control over sensitive information.

We believe applications should describe the intelligence they need while organisational policy determines where that computation is allowed to happen.

That's the idea behind Distributed Inference.

DI / TEAM / 09

Engineering-led.
Security-conscious.
Built for the real world.

Distributed Inference is being developed in Glasgow by a technical team focused on a difficult enterprise question: how can organisations use modern AI without surrendering control of sensitive information?

The work sits at the intersection of distributed systems, enterprise integration, AI infrastructure and information security. We approach it as critical infrastructure—not as another layer of AI theatre.

Editorial illustration of three technical colleagues collaborating around an infrastructure diagram in a Glasgow workspace.
TEAM / GLASGOWIllustrative image
01 / ENGINEERING

Infrastructure first

Clear interfaces, operational discipline and integration with the systems organisations already rely on.

02 / SECURITY

Controls are the product

Policy, minimum necessary disclosure, credential isolation and auditability are treated as core requirements.

03 / PARTNERSHIP

Built with practitioners

We want technical, security and privacy leaders to challenge the assumptions and shape useful pilot outcomes.

WHAT A PILOT ORGANISATION CAN EXPECT

Direct, technical collaboration

  • Direct access to technical decision-makers
  • Defined scope and shared success criteria
  • Architecture and security review
  • Documented assumptions, risks and constraints
  • Honest limits instead of inflated claims

DI / PILOT / 10

Explore a pilot
with us

We're looking for organisations that handle sensitive or regulated information and want to evaluate Distributed Inference through a controlled pilot.

We're particularly interested in speaking with technology, security, privacy and AI leaders across professional services, healthcare, life sciences, financial services, defence, the public sector, energy and research-intensive organisations.

STEP 1 OF 2

Tell us about you