Opengeni

What it does

Opengeni provides the infrastructure around AI agents, not the agent itself. Every session is stored as an event log in Postgres, can run for hours or days and can be traced step by step afterwards. The agent pauses risky tool calls until a person approves them. It reaches your product through an MCP server or an OpenAPI description; credentials are managed per user, and memory carries over between sessions. Work runs either in a managed sandbox or on a machine you enroll yourself. Scheduled tasks and webhooks start background work, and ready-made React components supply the chat interface.

Who it suits

Development teams and software vendors who want to build agents into their own application and need tenant isolation, approvals and traceability. Without programming skills it is of little practical use.

Data protection for businesses

The cloud version runs on Microsoft Azure in the EU. Sandboxes at the provider Modal, however, are not tied to a region and may run in the US; depending on the model you choose, content goes to OpenAI, xAI or other providers, and with DeepSeek also to China. According to the vendor, your content is not used to train models. You can request a data processing agreement from support. There are no dedicated business plans; self-hosting gives you full control over your data.

Limits

The project is still young, and the cloud service comes without an uptime guarantee. Self-hosting requires Kubernetes, Postgres, NATS and Temporal.