October
Request a demo

Every requirement. Traced.

October’s standing disclosure of large language model provision for offerors, prime contractors and contracting officers evaluating October in a US federal procurement.

VERSION 1.0 · JULY 2026

Commercial models via governed APIs.

No provider training on October data. Human oversight for consequential outputs. Published safety and bias evaluation.

Disclosure index12 clause items
01

Acceptable use policy

Published
02

Model, system and data cards

Published
03

Architecture and data lineage

On request · NDA
04

Development artifacts

On request · NDA
05

Governance and evaluation artifacts

Published
06

Human-centred design approach

Published
07

Model merging or blending

Not applicable
08

Versioning and quality assurance

Published
09

Architecture and workflow diagrams

Published
10

End-user resources

Published
11

Feedback mechanisms

Published
12

Unbiased AI principles

Published
ArchitectureInput → safeguard → model → review

Models stay behind
controlled boundaries.

October embeds commercially hosted foundation models through zero-retention API configurations. Purpose, provider, version, input class, output use and oversight are registered.

Open the model register
01

User or system input

02

Policy and safety controls

03

Registered model endpoint

04

Output validation

05

Human or product action

Operating positionReviewed quarterly
01

Model provision

October uses commercially hosted OpenAI and Google models via API. The active models, deployment contexts and purposes are named in the public register.

02

Training and retention

Providers are contractually prohibited from using October customer data for training. Eligible API access uses zero data retention.

03

Consequential use

Hiring support remains advisory. Crisis and safeguarding signals break to a human or emergency pathway; the agent does not act beyond scope.

04

Testing and traceability

Prompt bias, flip testing, harm detection, concern recognition and safeguarding evaluations are published with method, sample and limitations.

05

Artifacts on request

Architecture decision records, detailed data lineage, threat models and development artifacts are maintained internally and can be shared with contracting agencies under NDA.

06

Accountability

The CEO is accountable for the programme; technical, privacy, product and governance owners maintain controls and change-triggered retesting.

Start with your organisation

Need the procurement pack?

Contact October’s Data Protection Officer for NDA-controlled technical and governance artifacts.