Bring governed portfolio data
into the AI systems you already use.

1. 01

   Approved enterprise AI client

   An assistant, internal agent, or analytics workflow your organization already operates.
   Request — A natural-language question or query is sent to the PDWare MCP endpoint.
3. PDWare governance boundary

   02

   PDWare MCP gateway

   Implements the Model Context Protocol — the entry point for every AI request.

   Session — The request enters PDWare's governed access path; nothing reaches data directly.

   03

   Authentication & authorization

   The caller is identified and mapped to a PDWare role before anything is evaluated.

   Identity — An agent acting for a user sees only what that user's role allows.

   04

   Permissions, use rights & business rules

   Each request is evaluated against the same guardrails as the PDWare application itself.

   Permission evaluation — Out-of-policy requests stop here; approved queries continue.

   05

   Portfolio & resource data

   Demand, capacity, scenarios, and financials — the data PDWare already governs.

   Data access — The approved query runs against governed portfolio data. No PDWare-side model training.
5. 06

   Governed response

   Results return to the AI client scoped to the caller's permissions.
   Response — The client uses the result for reporting, analysis, or an approved integration.

Every step inside the boundary is enforced by PDWare application security — the same permissions, use rights, and business rules that govern the application itself. For authentication method, audit logging, and validated clients, see Technical facts below.