Platform
The Enterprise AI Control Plane
One Control Plane for Every AI Model, Agent, Workflow, Tool, and Infrastructure Layer.
A self-hosted AI Security Gateway providing a single, unified, OpenAI-compatible interface to leading LLM providers and models, centralizing access, routing, security, cost control, and observability.
AI Gateway
Single OpenAI-compatible endpoint. Drop-in compatibility: change only the base URL, no code rewrite.
LLM Routing
Application-level load balancing. Automatic fallbacks and retries on provider outage/errors.
FinOps
Real-time spend tracking per key / team / user. Budgets and quotas with automatic threshold alerts.
Identity
Virtual keys, scoped, revocable, per app / project / team / user. SSO / SAML integration with RBAC.
MCP Governance
Unified endpoint for LLMs, agents, A2A, and MCP tools, governed within the tool-control framework.
GenAI Firewall
Every prompt and response inspected, policy-checked, and filtered in real time.
Audit
Audit logs for security and compliance. Usage analytics and traffic mirroring for A/B model evaluation.
Rate Limiting
Limits per user, team, app or agent stop loops, abuse and spikes before they are billed.
Caching & Batching
Repeated requests served from cache; non-urgent requests batched at lower provider rates.
Resource Pools & Model Routing
Each request goes to the most cost-effective model that fits the task. Capacity shared across teams.
Platform Architecture
Five integrated layers sit between your applications and the models they call, each enforcing a different dimension of control.
Gateway Layer
A single OpenAI-compatible endpoint fronts every provider and model. One base URL, no code rewrite, unified telemetry.
Security Layer
GenAI Firewall inspects every prompt and response in real time, with PII masking and data-loss prevention.
Governance Layer
Policy engine enforces per key and team rules. MCP tools, agents, and A2A flows governed under one framework.
Observability Layer
Full audit logs, usage analytics, and traffic mirroring for A/B model evaluation and incident review.
Cost Control Layer
FinOps, rate limiting, caching and batching, and model routing keep unit cost predictable and capped.
What the Control Plane Connects
Aidome sits between the things that consume AI and the infrastructure that serves it, applying policy and visibility across every layer.
Models & Providers
100+ providers and 2,000+ models behind one endpoint, with new models typically available within a day of release.
Agents & Tools
Agents, A2A flows, and MCP servers governed within the same tool-control framework as the models themselves.
Teams & Workspaces
Virtual keys, roles, and workspaces scope access per app, project, team, or user with RBAC and SSO / SAML.
Enterprise Systems
Connects to your identity, SIEM, and observability stack so AI usage is governed inside your existing controls.