Frequently asked questions

Praesidia is an AI infrastructure security platform. It authenticates, governs, and monitors every interaction between applications, agents, and MCP servers. Think of it as a purpose-built service mesh for AI infrastructure with bidirectional controls.

Traditional IAM providers think in terms of identities and permissions on individual resources. Praesidia thinks in terms of relationships between AI entities, with content-aware guardrails. Security lives at the connection between entities, not on the entities themselves.

Praesidia supports three entity types: Applications (web and mobile apps), MCP Servers (Model Context Protocol servers), and Agents (external agents or agents built and hosted through the platform). All three are first-class citizens with full credential management, guardrails, and policies.

Guardrails are content-level controls that define what can be said or requested in messages between entities. Policies are operational controls like rate limits, geo-restrictions, volume caps, time-based access, and IP allowlists. Both can be configured per entity, per direction (as a client or as a server).

A connection defines a trusted relationship between two entities. It specifies which entities can communicate and what controls (guardrails and policies) govern that communication. Connections can be simple point-to-point or orchestrated multi-layered setups.

Each entity registers with Praesidia and receives credentials (client key + secret). Every request between connected entities is authenticated through Praesidia using OAuth2-based credential management. This ensures only authorized entities can communicate.

Yes. During the beta phase, Praesidia is completely free for all users. There is also an open-source version for individual developers. No credit card required to get started.

Yes. The core components are Apache-licensed and open source. Commercial SaaS and enterprise packages will be available on top of the open-source core for teams that need additional features.

Sign up at app.praesidia.ai, register your first entity (app, MCP server, or agent), and create a connection. You can have authenticated AI interactions running in minutes. Check the documentation for step-by-step guides.

Praesidia offers a Python SDK (install via pip install praesidia-server-sdk), CLI tools, and simple credential management. Integrations with frameworks like LangChain and CrewAI are planned.

Governance features including compliance frameworks (SOC2, ISO 27001, HIPAA, GDPR) and audit trails are coming soon. The platform is being built with compliance in mind from day one, guided by our CISO who led Romania's first ISO/IEC 42001 AI management certification.