Partners

The technology we build on.

We are model-agnostic and cloud-agnostic by policy: the stack is chosen for each project against its constraints, and it lives in your accounts, not ours. This page names the platforms we work with most, and how we work alongside other firms.

Platforms and models

These are tools we use and have shipped with, listed so you can see whether your existing estate fits. Naming them claims nothing more than that: we hold no certified, official or reseller partner status with any of them, and none endorses us. Where we recommend one over another, the reason is the project's constraints, never a commercial relationship.

Cloud & compute: Amazon Web Services Microsoft Azure Google Cloud Hetzner On-premise GPU servers
Model providers: Anthropic OpenAI Google Mistral Open-weight models
Retrieval & data: Vector databases PostgreSQL Elasticsearch Snowflake
Serving & observability: vLLM Docker Kubernetes Observability tooling

All names are the trademarks of their owners. Choices for a given project are documented in the proposal.

Working alongside other firms

Agencies and studios

We take on the AI engineering inside a larger project you are running for your client: retrieval, agents, integrations, evaluation. White-label or named, as your client prefers. Several of the campaigns and builds on our work pages came to us this way.

Systems integrators

If you deploy into a vertical and need a team that can design and build the model layer to a security review's standard, we scope it with you and hand over code your engineers can operate.

Your in-house team

Most engagements end with your engineers running the system. We plan for that from the first week, which is why the code lives in your repositories and the decisions are written down.