A full-stack enterprise AI provider, covering infrastructure, models, agents and operations, aligned with European data protection.
We run our own infrastructure in Spain and keep data in the EU. Companies use our platform to put AI agents to work: agents that call tools, follow complex instructions, read long documents and answer in the languages their teams speak.
| Product | What it does |
|---|---|
| Serenity Star | Agent orchestration: agents, knowledge (RAG), connectors and workflows, with a compliance centre and bring-your-own-model. |
| Serenity AI | For small and medium businesses: work with GenAI models, agents and microapps. |
| Serenity Edge | An OpenAI-compatible inference API on our own infrastructure in Spain, with data in the EU and zero data retention (the prompt cache lives for less than 24 hours). Edge traffic is never used to train, fine-tune or evaluate models. |
| Orion | Our model family, served through Serenity Edge and Serenity Star. |
| Model | For |
|---|---|
orion-pro |
Demanding tasks: long documents, complex instructions, multi-step tool calling. |
orion-plus |
Large volumes: chat, classification, extraction. |
| Orion Compass (public preview) | Questions that need depth. |
| Compass Flash (public preview) | Lower latency for high-volume routing, filtering and triage. |
The Orion family keeps evolving. Orion weights are not published here: the models are available through Serenity Edge and Serenity Star (model list).
Substrate Lab is the team behind Orion. Its work centres on enterprise agent workloads (tool calling, structured output and grounded reports) and on European languages beyond English.
Substrate Artificial Inteligence, S.A.