NHS and public sector health
AI inference that stays in the UK.
Private AWS infrastructure in London. No US sub-processors. No training on patient data.
Most AI API products process your data on infrastructure outside UK jurisdiction. Marigold runs open-weight models on private AWS infrastructure in London. No inference request leaves the UK. No external model provider receives your data. Built for organisations where data residency is not optional.
NHS trusts, CCGs, and public sector health organisations are adopting AI tooling and hitting the same wall: US-based providers cannot satisfy UK data residency requirements. Marigold is a UK company running UK infrastructure with no US model provider in the request path.
Data handling properties
01
UK data residency
Inference runs on private AWS infrastructure in London (UK). Data does not leave that region as part of inference, storage, or queuing. No cross-border transfer occurs.
02
No US sub-processors
No inference request reaches OpenAI, Anthropic, Google, or any other US-based model provider. The model runs locally from weights held on UK infrastructure.
03
No training on inputs
Inference inputs are not retained for training, fine-tuning, or any other secondary purpose. Outputs are stored briefly for retrieval and then deleted.
04
Open-weight models
Models run from publicly documented open-weight checkpoints. The weights, architecture, and behaviour of each model are inspectable. No proprietary black-box model receives your data.
05
Audit trail
Every inference job records status and timing within your region. Execution history is queryable. Access logs are retained in CloudWatch within the UK region.
06
UK-registered company
Bay Information Systems is a UK-registered company. Any contractual relationship is with a UK entity subject to UK law.
DSPT considerations
Designed with UK data residency requirements in mind
The NHS Data Security and Protection Toolkit requires that personal data is processed with documented controls and within appropriate jurisdictional boundaries. Marigold's architecture addresses the aspects of that requirement that relate to AI inference: where the data goes, who can access it, and whether it is retained.
We are not able to make a blanket claim that Marigold satisfies your organisation's DSPT obligations -- that assessment depends on your specific data types, use case, and information governance framework. What we can do is provide the technical detail your IG team needs to make that assessment, and discuss your requirements before any commitment is made.
| Property | Marigold | US-based AI API |
|---|---|---|
| Data processed in UK | Yes | No |
| UK-registered processor | Yes | No |
| No US sub-processor for inference | Yes | No |
| No training on inputs | Yes | Varies / opt-out required |
| Audit log within UK region | Yes | No |
| Inspectable open-weight models | Yes | No |
Use cases
These are illustrative. Whether any specific use case is appropriate for your organisation depends on your data classification and IG framework.
Clinical document summarisation
Summarise discharge letters, referral notes, and clinical correspondence. All processing within UK infrastructure. No patient data leaves the region.
Medical transcription and briefing
Convert clinical audio or text notes to structured summaries. Produce multilingual patient-facing documents from a single submission. Welsh language support included.
Document classification and routing
Classify incoming correspondence by category, urgency, or department. Route to the appropriate team without manual triage. No data leaves UK infrastructure.
Frequently asked
Does Marigold process data outside the UK?
No. Inference runs on private AWS infrastructure in London (UK) by default. Data does not leave that region as part of inference. No US-based model provider receives your data.
Does Marigold train on patient data or inference inputs?
No. Inference inputs are not retained for training, fine-tuning, or any other purpose. Outputs are stored briefly for retrieval and then deleted.
Is Marigold suitable for NHS DSPT requirements?
Marigold's architecture is designed to support UK data residency requirements. Whether it satisfies your organisation's specific DSPT obligations depends on your data types, use case, and your IG team's assessment. We are happy to discuss the technical architecture in detail to support that assessment before any commitment is made.
Is a data processing agreement available?
We are able to discuss data processing terms with organisations that have specific requirements. We do not currently have a standard DPA document ready to issue -- if you have requirements in this area, contact us and we will work through them with you.
Is Marigold on G-Cloud?
Not currently. If G-Cloud listing is a requirement for your procurement process, let us know -- understanding the demand will inform when we prioritise that registration.
What models are available?
The hosted registry includes instruct models for text generation and summarisation, embedding models for search and classification, text-to-speech in English, Welsh, and several other European languages, and image-to-text models. See the full registry.
AI inference built for UK data requirements.
If you are evaluating AI tooling for an NHS trust or public sector health organisation, we would like to hear from you. Register your interest below or contact us directly.
Register interest
Tell us your organisation type and use case and we will follow up directly.
Noted. We will be in touch.
Prefer to talk directly? ed@bayis.co.uk