How We Work
A plain-language specification of what we build, how we build it, and what you receive at handoff.
What We Build
We build AI systems that run entirely inside your own infrastructure. This means the models, the data pipelines, the APIs, and the user interfaces are all deployed on servers you own or control. Nothing is hosted by us, and nothing is routed through external AI providers.
The result is a system that your compliance team can audit, your IT team can maintain, and your organisation owns outright. There are no ongoing licence fees, no vendor lock-in, and no dependency on our continued involvement to keep the system running.
Target Architecture
Every engagement is scoped to your specific environment. We do not sell a pre-built product. The architecture we design will depend on your existing infrastructure, your compliance requirements, and the use case you are solving.
That said, most of our deployments share a common structure: a self-hosted language model or specialist AI model, a governed data pipeline that feeds it, an integration layer that connects it to your existing systems, and an access control layer that enforces your security policies.
We work with on-premise servers, private cloud environments (AWS GovCloud, Azure Government, private GCP), and air-gapped networks. We do not deploy to public cloud environments where data could be accessed by the cloud provider.
What You Receive at Handoff
At the end of every engagement, you receive a complete, documented, production-ready system. The handoff package includes:
Every line of code written during the engagement, committed to a repository you own, with clear commit history and inline documentation.
All infrastructure-as-code files, environment configurations, and deployment scripts. Your team can reproduce the deployment from scratch using these files.
A full technical specification of the system: architecture diagrams, data flow documentation, API references, and operational runbooks.
A structured handoff session with your technical team covering system operation, monitoring, common failure modes, and how to extend the system.
Documentation prepared for your compliance team: data flow diagrams, access control specifications, audit log configuration, and model cards for any AI components.
Timeline and Engagement Model
Our engagements run between 30 and 90 days, depending on scope. We work on a fixed-fee basis. The scope, deliverables, and price are agreed in writing before any work begins. There are no hourly rates, no change orders, and no surprise invoices.
We take on a limited number of engagements at any one time. This is intentional. It means the people who scope your project are the people who build it.
We do not take on engagements where the client is not ready to move to production. If you are still evaluating whether AI is right for your organisation, we are probably not the right partner yet. We work best with organisations that have a specific use case, a clear compliance requirement, and the internal commitment to see an implementation through.
Compliance Frameworks We Work Within
Our deployments are designed to operate within the following frameworks. We do not certify your organisation against these frameworks, but we build systems that are compatible with them and produce the evidence your compliance team needs.
Healthcare data handling, PHI protection, audit logging, and Business Associate Agreement compatibility.
Data minimisation, purpose limitation, right to erasure, and cross-border transfer restrictions.
Security, availability, and confidentiality controls. Audit trail generation for SOC 2 evidence collection.
Information security management system compatibility, risk assessment documentation, and control implementation evidence.
Ready to move past the pilot?
Tell us about your environment. We will respond within one business day.