experientiallabs/experiential

An open source model gateway that provides one control plane across closed, open-source, local, and custom models.

What it solves

Experiential is an open-source gateway and router for agent workflows. It simplifies the management of multiple LLM providers by providing a single OpenAI-compatible API, while adding a layer of control over user access, model selection, and spending limits. It also allows developers to turn production traffic (telemetry) into optimized routers or specialized models.

How it works

The system acts as a proxy between the agent and the LLM providers. It supports hosted, Bring Your Own Key (BYOK) and local models. Users can set up a local gateway using a setup wizard or use a managed hosted platform. The platform collects OpenTelemetry traces from agent traffic, which can then be used to build simulations and optimize a router or fine-tune an open-source model for better quality, speed, and cost.

Who it’s for

It is designed for developers building AI agents and those managing multiple LLM providers across different use cases, needing centralized control over costs and access.

Highlights

  • Unified API: Access hosted, BYOK, and local models through one OpenAI-compatible interface.
  • Control Plane: Manage which users and agents can use specific models and set spending budgets.
  • Traffic-to-Optimization: Convert production OpenTelemetry traces into custom routers or optimized models.
  • Flexible Deployment: Available as a local gateway or a managed hosted platform.

Related

  • Dispatch
  • Project
  • Project
  • Project
  • Project