Use any agent and any inference provider

Agent docs Talk to us

Run coding agents without being locked into one model

Bring your AI gateway or use ours. Proliferate is fully agent and inference provider agnostic.

Run Claude Code, Codex, Gemini, OpenCode, Cursor, and custom agents.
Route inference through provider keys, gateways, or local endpoints.

Agent docs

Agent and inference, controlled by you
Compatibility without losing governance

Agent registry

Define which agents teams can launch.

Bring your AI gateway or use ours

Compatible with Bedrock, LiteLLM, OpenAI, Anthropic, and more.

Custom LLM models

Proliferate can help you fine-tune open source models for your needs.

Limits & Budgets built in

Control per-user and per-org budgets for LLM usage.

Questions & answers

Do we have to use one agent?

No. The platform is built to support multiple agent runtimes and custom agents.

Can we use self-hosted models?

Yes. The model-routing layer is intended to work with provider keys, gateways, local endpoints, and self-hosted models.

Can admins limit model sources?

Yes. Source tiers and team policies are part of the governance layer.

Can different teams use different agents?

Yes, while still sharing the same environment, workflow, and policy system.

Other platform pieces
The rest of the agent infrastructure layer.

Integrations

Connect approved tools to agent runs.

Workflows

Schedule, trigger, and review agent runs.

Environments

Isolated workspaces with repo context.

Self-host Proliferate

Run Proliferate entirely in your VPC

Runtime Security

Control what agents can access and execute.

Governance & budgets

Policies, approvals, usage, and spend.

Security First

Identity, auditability, and deployment boundaries.

API, MCP, CLI

Coming soon

Programmatic access to agent workflows.

Run agents on infrastructure you control.

Start with one team, then standardize agent workflows without giving up control of code, credentials, or inference.

Download for Mac Talk to us