Content handling
Zero content retention
Prompts and responses are never written to logs, metrics or traces. Telemetry contains operational metadata only.
One endpoint for every model.
Control stays with you.
Connect provider accounts once. Give every team one production API, enforce access and spend centrally, and switch models without shipping application changes.
See every request.
Control every limit.
Usage, cost, latency and errors are attributed to the project, key and model that produced them. Hard budgets can stop over-limit requests before provider traffic begins.
Showing Overview
Why a gateway
Model choice should be a configuration decision.
Providers and models will keep changing. Aixy moves that change out of application code and into one control plane owned by your platform team.
Operating problem Without a control layer With Aixy
Provider changes
Without a control layerApplication teams update SDK configuration, credentials and model IDs.
With AixyPlatform teams change one central route. The application API stays stable.
Credential access
Without a control layerProvider secrets spread across repositories, environments and teams.
With AixyWrite-only encrypted credentials live at organization or project scope.
Spend control
Without a control layerOverspend becomes visible after provider invoices arrive.
With AixyHard budgets reject over-limit requests before they reach a provider.
Operational reporting
Without a control layerCost, latency and errors live in separate provider dashboards.
With AixyEvery request is attributed to its project, key, provider and model.
153 providers. One gateway.
Connect OpenAI, Anthropic, Gemini, Mistral and the rest of your AI stack once. Applications keep one stable API while your platform team controls credentials, model access and routing centrally.
Applications
Applications
Aixy·
Provider network
OpenAI 
Anthropic 
Gemini 
Mistral More providers
- One API contract
- Applications stay unchanged as models and providers change.
- Bring your own keys
- Scope provider credentials by organization or project.
- Central routing policy
- Select, restrict or replace routes without redeploying clients.
Enterprise controls
Answers for security, finance and platform teams.
Each control maps to a concrete operational requirement—data residency, identity, spend enforcement, organization-wide monitoring or model safety.
Residency & privacy
Frankfurt, end to end
Regional AWS services stay in eu-central-1, with data boundaries designed for GDPR-compliant deployments.
Identity & access
RBAC with enterprise SSO
Organization and project scopes, granular roles and project keys, with SAML 2.0 and OIDC for enterprise plans.
Spend enforcement
Budgets that stop requests
Set daily, weekly, monthly or lifetime limits across organizations, projects, teams, users and API keys.
Advanced monitoring
See the whole organization
Track and export requests, spend, tokens, errors and p95 latency across every team, key, provider and model. Deliver incidents to email, Slack, Discord or webhooks.
LLM guardrails
Block unsafe input locally
Detect prompt injection, jailbreaks and sensitive data before opening a provider connection. Block or redact by organization or project without sending prompts to an external classifier.
Start with a real request
Put one stable endpoint in front of your LLM stack.
Connect a provider credential, create a project and route your first request. Talk directly to the team building Aixy when enterprise identity or procurement enters the conversation.