New from the Lab·The Compass — an open moral reasoning standard for AI, tested across frontier modelsExplore →
Production AI Institute · PSF v1.1 open standard
AI Right-To-KnowAI Data Use IndexCheck My AI ToolsPolicy Change WatchAgent ReadinessPublic BenchmarkContactGlobal standard · Worldwide
Pattern Library21 Patterns · PSF-mapped · PAI-8 aligned

Agentic Design Patterns
A Production AI Reference

The 21 architectural patterns that underpin every production AI system — each mapped to PSF safety domains and PAI-8 governance controls, with enterprise examples, production failure modes, and implementation checklists. Written for practitioners who need to deploy, govern, and audit these systems.

Read the PSF →Open PAI Studio
Build it, then prove it

Patterns are not the outcome. Safer deployments are.

Once a pattern maps to your system or client conversation, move into the path that creates proof, revenue, or a working design.

Part 1

Core Patterns

The seven foundational patterns every production AI practitioner needs to understand. These appear in virtually every enterprise AI system.

PSF D2PSF D1PAI-8 C5

Prompt Chaining

Sequential task decomposition where each model output feeds the next input.

PSF D1PSF D5PAI-8 C1

Routing

A classifier that directs each input to the most appropriate specialist agent or pipeline.

PSF D4PSF D5PAI-8 C6

Parallelism

Running multiple agent tasks simultaneously and synthesising the results.

PSF D2PSF D6PAI-8 C5

Reflection

An agent critiques and revises its own output before it reaches a human.

PSF D1PSF D7PAI-8 C2

Tool Calling

The pattern that turns a language model from a text generator into an actor.

PSF D6PSF D5PAI-8 C4

Multi-Agent Collaboration

Specialised agents working together, each owning a domain of the overall task.

PSF D6PSF D4PAI-8 C1

Orchestration

A controlling agent that directs sub-agents, manages state, and decides when a task is complete.

Part 2

Production Patterns

Seven patterns that determine whether a system survives contact with real users, real data, and real operational environments.

PSF D3PSF D4PAI-8 C3

Memory Management

How agents store and retrieve information across sessions, tools, and agent boundaries.

PSF D5PSF D7PAI-8 C6

Exception Recovery

How agents detect failure and decide whether to retry, escalate, skip, or fail gracefully.

PSF D6PSF D2PAI-8 C4

Human-in-the-Loop

The architecture for deciding when agents act autonomously and when they pause for human review.

PSF D1PSF D2PAI-8 C5

Safety Guardrails

The input and output filters that prevent agents from receiving or producing content they should not.

PSF D4PSF D2PAI-8 C7

Performance Evaluation

Systematic measurement of whether agents produce the right outputs at the right quality level.

PSF D3PSF D2PAI-8 C3

Context Window Management

Strategies for fitting the right information into the finite context an agent can process.

PSF D3PSF D1PAI-8 C3

Retrieval-Augmented Generation

Connecting agents to external knowledge so they can retrieve facts rather than hallucinate them.

Part 3

Enterprise Patterns

Seven patterns for complex, high-scale, high-stakes AI systems operating at the frontier of what production AI can do today.

PSF D5PSF D4PAI-8 C6

Event-Driven Agents

Agents triggered by events in your systems rather than by direct user prompts.

PSF D4PSF D2PAI-8 C1

Feedback Loops

Architectures that route agent outputs back as inputs to improve the next cycle.

PSF D6PSF D4PAI-8 C4

Swarm Intelligence

Many simple agents working in parallel on variations of a problem, synthesised into one output.

PSF D6PSF D5PAI-8 C1

Hierarchical Agents

An agent hierarchy where strategic agents direct tactical agents that direct operational agents.

PSF D7PSF D6PAI-8 C4

Self-Improving Agents

Agents that propose improvements to their own configuration — with mandatory human approval.

PSF D2PSF D6PAI-8 C5

Debate and Verification

Two agents take opposing positions; a third evaluates the debate and produces a verified conclusion.

PSF D4PSF D2PAI-8 C7

Curriculum Learning

Agents tested against progressively harder evaluation sets, with difficulty dynamically adjusted on performance.

Apply these patterns

Use the pattern library as an implementation reference for PSF-aligned systems. Credentials are available when you need individual or organisational proof against the same standard.

Read the PSFOpen StudioView credentials
The Production AI Brief

Get the brief that keeps AI work defensible

PSF updates, deployment checks, failure patterns, and proof paths for practitioners, MSPs, and teams who need AI work to survive scrutiny. No hype.