- Organisation
- OneAdvanced, United Kingdom
- Predecessor
- Initial Amazon Bedrock prototype for the AI use cases described in the cited AWS evidence during a two-week sprint
- Destination
- UK-sovereign production AI architecture in AWS London using Llama 4 on Amazon SageMaker AI, more than 50 Strands agents on Amazon ECS and supporting AWS storage/database services
- Workload
- Amazon SageMaker AI, Amazon ECS and AWS UK-sovereign AI architecture
- Stage
- operational
Verified platform change
AWS and OneAdvanced reported on 12 August 2026 that OneAdvanced operates a UK-sovereign production AI architecture in AWS London, self-hosting Llama 4 Maverick and Llama Guard 4 on SageMaker AI and running more than 50 Strands agents on ECS. The source states the solution has been running in production since July 2025.
Transition
From: Initial Amazon Bedrock prototype for the AI use cases described in the cited AWS evidence during a two-week sprint
To: UK-sovereign production AI architecture in AWS London using Llama 4 on Amazon SageMaker AI, more than 50 Strands agents on Amazon ECS and supporting AWS storage/database services
Known unknowns
The dated AWS/OneAdvanced source proves the described UK production architecture by 12 August 2026 and states that data does not leave the UK. It does not establish 12 August as a private cutover day, claim a whole-company infrastructure migration, or prove retirement of every predecessor platform.
Official sources
Updates
Source rechecked on 26 September 2026. The current first-party AWS article co-authored with OneAdvanced has not materially changed; the bounded UK-sovereign production AI architecture and operational state remain supported without a private exact-cutover, whole-company migration or every-predecessor-retirement claim.