liveBook · Manning

Architectural style

As explained in Architecting for Autonomy

A recurring architectural blueprint describing a solution structure, component interactions, and architectural constraints.

An architectural style is a recurring blueprint for structuring components, defining their interactions, and enforcing constraints. In agentic systems, it connects agents with data, runtime policies, existing platforms, and supervision. section 1.7

A useful architectural style makes three design boundaries explicit:

  • Interaction boundary: specifies how agents interact with data and platforms. section 1.7
  • Authority boundary: evaluates actions against policy-as-code before execution. section 1.7
  • Failure boundary: defines whether unavailable policy evaluation fails closed or routes the action to human review. section 1.7

Policy Gateway

The Policy Gateway pattern places a policy-as-code decision before every agent action. It consistently enforces authority and constraints, while specifying a controlled response when the policy engine is unavailable. section 1.7

Figure 1.7 The Agentic Enterprise Framework organizes the architect’s work into three actionable streams: patterns, playbooks, and practice, moving from theoretical design to a governed, operational Foundation for Autonomy.

Use the framework view to distinguish architectural patterns from implementation playbooks and governance practice. section 1.7

How Architecting for Autonomy covers it

How deep it goes
Defines recurring blueprints and failure boundaries but leaves implementation verification to playbooks and day-to-day control to practice.
Examples
Uses the Policy Gateway, pre-execution policy evaluation, fail-closed behavior, and routing to human review.
What it assumes
Assumes architectural rigor but does not require AI-engineering expertise; teaches runtime boundaries, policy-as-code, and continuous supervision.
What you can do after
Evaluate/understand how autonomous actions receive explicit authority, constraints, supervision, and failure handling.
How it connects
Connects agents, data, policy-as-code, runtime guardrails, human review, platforms, playbooks, and governance practice.

Also explained in

Compare how 3 books explain architectural style →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Architecting for Autonomy. The section links point to the parts of the book it draws on.