Security & Deployment
Run Nautili where your data lives and pass your security review with confidence.
Nautili gives you a secure, controllable way to route AI traffic across providers and models. Deploy in your environment or use our managed cloud. You keep control of your data, keys, access and policies.
Where it runs
Deploy Nautili in the environment that works for you. You manage what you host.
Shared Gateway
managed by Nautili
Use a shared Nautili-managed gateway with a static URL and Nautili-issued API keys. Traffic is attributed to the team associated with each key.
Self-hosted
your environment
same nautili controls in every mode
routing · policy · observability
Data flow
What passes through Nautili, what is retained, and what is not.
Passes through
Customer Content: prompts, inputs, files, messages, completions and outputs. Model Providers process it under their own terms and privacy policies.
Customer Content is not retained
Processed only as needed to route the request, return the response and operate the platform. Not retained after the request completes, except transiently in memory or short-lived processing buffers.
Request metadata is retained
API Data such as model, Model Provider, token usage, request and response size, latency, timestamps, status, cost and the API key or Account used, for observability, routing, rate limiting, billing and support. API Data does not include Customer Content.
Keys and providers
Bring your own provider keys or GPUs. Keys stay in your deployment.
Your provider keys
Use your own API keys for OpenAI, Anthropic, Google, AWS, Azure and any OpenAI-compatible provider.
- openai
- anthropic
- aws
- azure
- openai-compatible
Your own models and GPUs
Route to self-hosted models in your environment, including on your own GPUs.
- self-hosted model
- your GPUs
- your network
Access control
Control which models and providers each team or key can use.
Team permissions
Allow or restrict providers and models by team, environment, or API key.
Scope
- team
- environment
- api key
TeamResearch
{ "team": "research", "allowed_providers": ["openai", "anthropic"], "allowed_models": ["gpt-4o", "claude-3-5-sonnet"], "rate_limits": { "rpm": 100, "monthly_usd": 500 }}Policy · applied at the gateway on every request
Audit
A request record for every response, so you can see what happened.
- OpenAI / gpt-4oSuccess2026-10-08 14:32:07 · research
- Latency
- 842 ms
- Cost (USD)
- $0.0124
- Request ID
- req_7f3a9c21e4b8
- Anthropic / claude-3-5-sonnetSuccess2026-10-08 14:31:52 · support
- Latency
- 1,214 ms
- Cost (USD)
- $0.0310
- Request ID
- req_2b81d07c55aa
- AWS Bedrock / llama-3-70bError2026-10-08 14:31:18 · research
- Latency
- 2,031 ms
- Cost (USD)
- $0.0000
- Request ID
- req_91aa3e7d0c46
- OpenAI / gpt-4oSuccess2026-10-08 14:30:59 · platform
- Latency
- 788 ms
- Cost (USD)
- $0.0102
- Request ID
- req_5d2c8b9e1f70
Each response carries its request ID, so a record can always be traced.
Compliance
We’re happy to answer your security and compliance questions.
Our current certifications, security documentation and terms are available on request. Please contact the Nautili team for the latest information.
For security and procurement review
- Security overview
- Available on request
- Deployment architecture
- Available on request
- Data handling
- Available on request
- Terms
- Available on request
- Contact
- Talk to the team
Deploy Nautili with confidence
Talk to our team about your security requirements or explore the deployment documentation.