Industries · Manufacturing

Engineering knowledge, locked down.

AI for design, manufacturing, quality and supply chain teams whose specs, standards and qualification records can’t leave the building.

Why it matters

Your governance process is right.

Manufacturers run on specs, standards, process records, supplier history and non-conformance data. That knowledge is the product, and much of it is bound by customer contracts or export rules.

Northwood deploys Oak inside your perimeter, with outbound traffic restricted to an allowlist and access flowing from your identity provider.

Use cases

Where manufacturers put Oak to work.

Standards reviewerCross-reference proposed design changes against your internal engineering standards and prior designs, and report the deltas.
Process spec assistantAnswer questions about process specifications, heat-treats, NDT requirements and qualification records, citing the controlling spec.
Supplier risk summaryPull supplier history from procurement records and produce a risk summary for sourcing reviews, inside your tenancy.
Non-conformance triageTriage non-conformance reports against prior dispositions, standards and specifications, and route them to the right owner with context attached.
Controls

What stays true in every deployment.

Outbound traffic is restricted to an allowlist. Off-list calls fail closed, logged and alerted.
Authentication and resource scoping flow from your identity provider. No platform-only accounts.
Northwood staff don’t have ambient access. Support sessions are time-boxed, named and logged on your side.
Deployment

Your cloud, your servers, or no connection at all.

See deployment options
Your cloudOak runs in your own cloud account, next to the data it reads.
Your serversRun Oak on hardware you own, in your data centre or your office.
Fully air-gappedNo outside connection. Models, search and logs all run locally.
Security

You don’t have to take our word for it.

Every deployment passes the AI Agent Security Audit before go-live, after every change and every quarter, and the results live in a repository you own.

10tests on the live system
  • Before go-live
  • After every change
  • Every quarter

Your knowledge, your models, your AI deployment.

Request a demoExplore Oak