# Doorman

> Doorman is an open-source (Apache 2.0), self-hosted AI and API gateway written in Rust. It gives each employee a personal AI key with an allowance in US dollars, tokens or requests, reports AI spend by person, team, model and provider, redacts secrets and personal data from prompts, and secures REST, GraphQL, gRPC and SOAP APIs with users, roles, rate limits, quotas and logs. It runs as one container in your own infrastructure.

Key facts:

- Price: free to self-host; every feature is in the open-source repository. Optional paid, assisted pilots.
- AI providers: OpenAI, Anthropic, Google Gemini, Azure OpenAI, and OpenAI-compatible servers. Doorman does not resell AI; providers bill you directly.
- Allowances: per person, group, tier, AI API or company; per minute, hour, day, week or month (UTC); block (HTTP 429, error AI004), warn, or downgrade to a cheaper model. They are soft limits.
- Guardrails: emails, US Social Security numbers, credit cards, AWS keys, private keys, API keys and custom patterns; redact, block or log.
- Storage: in memory with encrypted snapshots for one node, or MongoDB and Redis for several.
- Not supported today: files, batches, fine-tuning, realtime, assistants, image generation, and native Vertex AI and Bedrock.

## Product

- [Home](https://doorman.dev/): Overview: what Doorman is, who it is for, and how the AI and API gateway fit together.
- [AI governance](https://doorman.dev/ai-governance): The AI gateway: personal keys, allowances, guardrails, usage reporting, and its documented limits.
- [API management](https://doorman.dev/api-management): The API gateway: protocols, auth, rate limits, quotas, versioning, routing and logs.
- [Pricing](https://doorman.dev/pricing): Pricing: free self-hosting under Apache 2.0, and assisted pilots.

## Resources

- [FAQ](https://doorman.dev/faq): Frequently asked questions with direct answers.

## Compare

- [Compare](https://doorman.dev/compare): Index of comparisons and guides.
- [Doorman vs LiteLLM](https://doorman.dev/doorman-vs-litellm): Comparison with the LiteLLM proxy, including where LiteLLM is the better fit.
- [Doorman vs Kong](https://doorman.dev/doorman-vs-kong): Comparison with Kong Gateway, including where Kong is the better fit.

## Guides

- [AI spending limits per employee](https://doorman.dev/ai-spending-limits): Guide: per-employee AI spending limits in dollars, tokens or requests.
- [OpenAI proxy with per-user keys](https://doorman.dev/openai-proxy-per-user-keys): Guide: replacing a shared OpenAI key with per-user keys behind a self-hosted proxy.
- [Redact secrets from AI prompts](https://doorman.dev/ai-prompt-redaction): Guide: redacting secrets and personal data from prompts with gateway guardrails.

## Company

- [Contact](https://doorman.dev/contact): Contact form and pilot requests.

## Docs

- [README and quick start](https://github.com/doorman-dev/doorman#readme): install, demo and configuration.
- [AI gateway guide](https://github.com/doorman-dev/doorman/blob/main/user-docs/08-ai-gateway.md): providers, keys, allowances, guardrails, pricing and limits.
- [Operations](https://github.com/doorman-dev/doorman/blob/main/user-docs/05-operations.md): backup, restore, upgrades and key rotation.

## Optional

- [Full text of this site](https://doorman.dev/llms-full.txt): every page above as plain text.
- [Source code](https://github.com/doorman-dev/doorman)