Skip to content

The BPM/Agent Stack

Execution governance for AI agent architectures

Process models. Decision tables. Activity attributes. Structured exception handling. The proven discipline of Business Process Management — applied to the execution governance gap in AI agent deployment.

Every AI agent framework solves mechanism: how models access tools, how agents coordinate, how state persists. None solves execution governance: who is responsible for each step, what structured logic governs decisions, what happens when things go wrong, what constraints apply at each activity, and what audit trail exists.

The BPM/Agent Stack covers execution governance — orchestration, integration, and execution of authorized work. The companion Intent Stack specification covers governance context — from intent discovery through runtime alignment. Together, the two specifications address seven governance concerns across the full governance lifecycle.

The BPM/Agent Stack — 3-Concern Execution Governance Model showing Orchestration, Integration, and Execution layers, with Intent Stack governance context above connected through the stitching mechanism, and Constitutional AI substrate beneath.
The BPM/Agent Stack — Execution Governance Model

Who this is for

  • CISOs & security architects — governing AI agents performing security operations with SP 800-53 compliant execution infrastructure
  • BPM & governance practitioners — extending proven process, decision, and case management patterns to AI agent execution
  • AI agent developers — adding structured accountability, deterministic decision separation, and governed exception handling to agent architectures
  • Enterprise architects — deploying AI agents at scale with the same execution governance rigor applied to human workflows

Public Draft Specification, Version 1.1 — April 1, 2026. Grounded in BPMN 2.0, DMN 1.0, and the ABPMP BPM CBOK. Licensed under CC BY 4.0.