Poolside develops AI models and tools for software development. According to the vendor, it builds its own foundation models for agentic coding, such as the Laguna family, and offers them through a command-line tool called pool, a desktop application and an OpenAI-compatible API. Laguna S 2.1 has been released as an open-weights model, and the documentation shows how to run open-weight Poolside models locally with tools such as Ollama.
Poolside's documentation describes deployment on infrastructure you control: on-premises on a GPU host, or in a Kubernetes cluster on Amazon EKS, Red Hat OpenShift or upstream Kubernetes. Inference then runs in the customer's own environment. The company has a branch in Paris. The documentation does not say where the Poolside API runs, and the data processing agreement allows transfers outside the EEA, the UK and Switzerland where needed to provide the services. A European organisation should check which entity it contracts with and where the API and subprocessors operate, or choose on-premises deployment.