FOHS
FOHS provides a sovereign AI inference platform that runs models on UK data-centre GPUs while generating signed provenance receipts for every response. The service uses an OpenAI-compatible API, so existing applications can switch by changing a single base URL. Each receipt can be independently verified at fohs.ai/verify without an account.
- Artificial Intelligence
- Developer Tools
- Software Only
Funding
Founders
Product
Problem
Organisations running production AI workloads increasingly need to demonstrate where their data is processed, particularly when operating under UK jurisdiction and data-protection regulations. Standard cloud AI services often lack transparent, verifiable evidence of the geographic location and infrastructure used for inference, creating compliance challenges for data protection officers and legal teams.
Solution
FOHS provides a sovereign inference platform that runs AI models on UK data-centre GPUs and issues a signed provenance receipt for every API response. The service exposes an OpenAI-compatible endpoint, so developers can migrate by changing a single base URL to api.fohs.ai/v1 while keeping the same request shape, streaming behaviour, and SDKs. Each receipt contains operator, site, and signing-key details, enabling anyone to independently verify the response's origin at fohs.ai/verify without an account. The platform autoscales from zero to N replicas, allowing organisations to pay only for actual GPU-minutes of serving while maintaining full audit evidence for compliance reviews.
Target Audience
Primary customers are UK-based organisations running production AI workloads that need sovereign data processing and verifiable provenance evidence, including engineering teams and data protection officers responsible for regulatory compliance.
Features
- OpenAI-compatible inference endpoint requiring only a base-URL change for migration
- Signed provenance receipts attached to every response, containing operator, site, and signing-key metadata
- Public verification tool at fohs.ai/verify that requires no account to confirm receipt authenticity
- UK data-centre GPU infrastructure with autoscaling from 0 to N replicas and scale-to-zero capability
- Per-GPU-minute billing model that charges only for actual serving time