Agent Checkpoint and Recovery
The Agent Checkpoint and Recovery Pattern defines the durable execution architecture that…
Pattern Library
117 patterns across 13 architecture domains. Filter by category, maturity, or regulatory framework.
The Agent Checkpoint and Recovery Pattern defines the durable execution architecture that…
The Agent Identity and Authorisation Pattern defines the security architecture governing who…
The Agent Sandboxing Pattern defines the secure execution environment that isolates AI…
The Agent Tool Registry Pattern defines the centralised catalogue and invocation framework…
Agentic AI systems that autonomously execute multi-step workflows—calling tools, querying APIs, writing…
The Event-Driven Agent Pattern defines the architecture for AI agents that are…
The Long-Running Agent Pattern defines the architecture for AI agents that execute…
The Reflexive Agent Pattern defines an architecture in which an AI agent…
The Single Agent Pattern defines the canonical architecture for a single autonomous…
The Stateful Agent Memory Pattern defines the multi-tier memory architecture that allows…
APRA Prudential Standard CPS230 Operational Risk Management (effective 1 July 2025) imposes…
Australian Prudential Regulation Authority (APRA) Prudential Standard CPS234 Information Security applies in…
The EU Artificial Intelligence Act (Regulation (EU) 2024/1689) establishes legally binding obligations…
AI systems that ingest, process, or generate personal information must comply with…
ISO/IEC 42001:2023 is the first international standard providing requirements for an Artificial…
The NIST AI Risk Management Framework 1.0 (NIST AI RMF, published January…
AI systems introduce new vectors for cross-border data transfer that existing data…
The General Data Protection Regulation (Regulation (EU) 2016/679) applies in full to…
This page aims to provide an overview of the EU AI Act’s…
On 18 July 2025, the European Commission published draft Guidelines clarifying key…
The Code of Practice offers a clear framework to help developers of…
If your business uses AI to screen, rank, or match candidates, the…
Enterprises adopting Data Mesh architectures face a structural challenge when integrating AI:…
AI training data governance is the foundation of responsible AI.
Regulators, auditors, and risk committees increasingly demand that organisations explain not just…
AI systems are uniquely sensitive to data quality failures in ways that…
Federated learning enables multiple organisations or devices to collaboratively train a shared…
Privacy-by-design mandates that privacy controls are embedded into AI data pipelines at…
AI models that make real-time decisions — fraud detection, personalised recommendations, credit…
Many of the most valuable enterprise AI use cases — fraud detection,…
The AI Approval Workflow establishes a multi-stage, evidence-gated approval process that every…
The AI Audit Trail pattern establishes an immutable, tamper-evident log of every…
The AI Ethics Review Board (AERB) is a standing organisational governance structure…
The AI Incident Management pattern implements a specialised incident management lifecycle for…
The AI Model Register is the foundational governance artefact for any enterprise…
The AI Policy Enforcement pattern implements runtime governance controls that ensure AI…
The AI Regulatory Compliance Gateway is a centralised, runtime compliance validation layer…
The AI Risk Assessment Framework provides a systematic, repeatable methodology for evaluating…
The Model Bias Detection pattern implements a continuous pipeline for detecting, measuring,…
NIST has expanded its collaboration with the nonprofit MITRE Corporation as part…
This content is outdated – Draft guidelines have now been published by…
The Responsible AI Framework operationalises the five core responsible AI principles—fairness, transparency,…
The European Commission has published a call for applications for a scientific…
The Active Learning Loop pattern establishes a structured, closed-loop process by which…
The AI Confidence Threshold Routing pattern dynamically routes inference requests to different…
The Annotation and Feedback Loop pattern defines the end-to-end architecture for collecting…
The Collaborative AI Decision pattern defines an architecture in which humans and…
High-stakes AI systems — those authorising transactions, making clinical recommendations, or influencing…
The Human Escalation Pattern defines the architecture for routing AI-handled requests to…
The Human Override Pattern provides a lightweight, universally applicable mechanism by which…
The Hybrid Intelligence Pattern defines an architecture for systematically allocating each component…
The Enterprise AI Service Bus pattern establishes an event-driven integration backbone that…
The Legacy System AI Augmentation pattern adds AI capabilities to existing enterprise…
AI-Powered API Composition uses large language models with function-calling capability to dynamically…
Real-Time AI Stream Processing applies AI inference directly to streaming data pipelines,…
Batch AI Processing applies AI inference to large volumes of data through…
The AI Webhook Pattern delivers AI-generated notifications, inference results, and asynchronous AI…
An Enterprise Knowledge Graph (EKG) is a persistent, curated, machine-readable representation of…
The Semantic Data Layer (SDL) is a governed translation layer that sits…
The AI Knowledge Corpus Management pattern defines the complete operational lifecycle for…
Vector databases are the retrieval backbone of enterprise AI systems.
The Knowledge Graph for Explainability pattern uses a structured knowledge graph to…
Corpus Quality Assurance (CQA) is the automated pipeline that evaluates the fitness…
AI models are long-lived, continuously evolving artefacts.
Shadow model deployment allows an organisation to validate a new AI model…
Canary model release is the controlled, incremental exposure of a new AI…
Model rollback is the capability to revert production AI model serving to…
Multi-model ensemble combines the outputs of two or more AI models to…
Fine-tuning is the process of adapting a pre-trained foundation model to an…
Model compression reduces the computational footprint of an AI model — memory…
Model access governance controls who can invoke, export, copy, and administer AI…
AI systems present unique observability challenges that traditional APM tooling does not…
Prompts sent to large language models in production are the primary control…
Large language models fabricate plausible-sounding content with confidence.
AI system failures are qualitatively different from traditional software failures.
AI models degrade silently.
AI inference costs have a fundamentally different cost structure than traditional compute:…
AI pipelines are not single API calls.
AI system quality degrades silently between benchmarking events.
The AI API Gateway pattern establishes a purpose-built control plane that sits…
AI systems that cannot demonstrate systematic evaluation before production deployment are a…
Enterprise AI capabilities—LLM APIs, embedding services, vector stores, AI agent frameworks, fine-tuning…
The Enterprise AI Platform pattern establishes a shared, governed infrastructure layer that…
Feature stores solve a deceptively simple problem: when an ML model needs…
LLM inference is expensive and often redundant.
LLM inference costs exhibit a dangerous property shared with no previous enterprise…
The Model Routing pattern establishes intelligent, policy-driven dispatch of AI inference requests…
When an AI platform serves multiple tenants—whether internal business units within an…
Prompts are the primary programming interface for LLM-based systems, yet most organisations…
Agentic RAG places a reasoning AI agent in the orchestration loop of…
Contextual RAG with Metadata Filtering extends the foundational RAG pattern with a…
Retrieval-Augmented Generation (RAG) is the foundational architecture pattern that grounds Large Language…
Federated RAG enables retrieval across multiple organisationally or geographically distributed knowledge bases…
Graph RAG combines vector-based semantic retrieval with structured knowledge graph traversal to…
Hybrid RAG combines dense (semantic) vector retrieval with sparse (keyword-based BM25) retrieval…
Enterprise knowledge is never stored in a single system.
Multimodal RAG extends the retrieval-augmented generation paradigm to knowledge corpora that include…
Secure RAG is the enterprise access-control overlay that must be applied to…
Streaming RAG extends the foundational RAG architecture to continuously ingest and index…
Adversarial Input Defence addresses a class of attacks specifically designed to manipulate…
AI Data Classification defines the automated detection, labelling, and enforcement architecture that…
The AI Gateway pattern establishes a centralised, enterprise-grade control plane through which…
AI Output Filtering is the post-generation inspection and transformation pipeline that evaluates…
LLM Input Sanitisation is a pre-processing pipeline that transforms raw application inputs…
Model Isolation defines the architectural pattern for constraining the execution environment of…
The Prompt Firewall is an inline defensive layer that inspects every user…
Secrets Management for AI addresses one of the most prevalent and consequential…
Secure Tool Invocation defines the security architecture for AI agents that can…
The Zero-Trust AI Pipeline applies the "never trust, always verify" security architecture…