Patterns

A reusable architecture catalogue of patterns and heuristics for agentic AI systems. These patterns are derived from evidence across our experiments and learnings.

stable Apr 2026

Parallel Agent Execution

Concurrently executes independent agents or pipeline steps using async orchestration to reduce end-to-end latency without compromising state integrity.

Confidence: high
patterns concurrency performance
stable Apr 2026

Typed Execution Context

Replaces implicit dictionary-based state passing with a structured, versioned, and strongly typed object to ensure contract safety across pipeline steps.

Confidence: high
patterns state-management type-safety
stable Apr 2026

Active-Week Selective Intelligence

The merge limits full intelligence generation to active or explicitly requested weeks and uses deterministic fallback for non-active historical weeks.

Confidence: medium
agentic-ai active-week historical-data
stable Apr 2026

Cache-Aside Week Intelligence Builder

The merge introduces a cache-aside path for week intelligence so dormant weeks are materialized only when the UI requests them.

Confidence: high
agentic-ai cache-aside lazy-build
stable Apr 2026

Deterministic Brief Degradation

The merge introduces deterministic brief fallback when there is insufficient comparison context or intelligence is intentionally disabled for the week.

Confidence: high
agentic-ai fallback brief
stable Apr 2026

DTO-Backed UI Contract

The merge formalizes a response DTO so the UI reads a controlled output contract rather than arbitrary orchestration state.

Confidence: high
agentic-ai dto contract
stable Apr 2026

Dirty-Check Gated Lazy Recompute

The merge adds a deterministic dirtiness guard so week intelligence is recomputed only when stable training signals change or on-demand rebuild is forced.

Confidence: high
agentic-ai lazy-intelligence dirty-check
stable Apr 2026

Intelligence Snapshot Projection Boundary

The merge introduces a distinct derived model that aggregates agent and service outputs into a stable UI-facing intelligence snapshot.

Confidence: high
agentic-ai snapshot boundary
stable Apr 2026

Language-Keyed Intelligence Cache

The merge uses `(week_start, language)` as the intelligence cache key so multilingual outputs are reused safely.

Confidence: high
agentic-ai cache i18n
stable Apr 2026

Persisted Chart Artifact Pipeline

The merge converts visualization outputs into serialized image artefacts that can be persisted, reloaded, and rendered without live figure state.

Confidence: high
agentic-ai visualization persistence
stable Apr 2026

Scope-Aware Visualization Context

The merge adds explicit visualization scope so agents and rendering logic know whether they are producing week-scoped or full-run charts.

Confidence: medium
agentic-ai visualization-scope context-shaping
stable Apr 2026

Snapshot ID-Based Partial Enrichment

The merge preserves weekly snapshot identity so later intelligence and chart enrichment can target a stable persisted record.

Confidence: high
agentic-ai identity partial-update
stable Apr 2026

Week-Scoped Repository Reconstruction

The merge reconstructs intelligence and visualization contexts from repository-backed week state instead of relying only on newly uploaded runs.

Confidence: high
agentic-ai reconstruction week-scope
exploring Mar 2026

Pipeline Intelligence Snapshot

Produce a canonical intelligence artifact at the end of a pipeline to consolidate distributed agent outputs into a single system state.

Confidence: medium
patterns agentic-ai pipeline
exploring Mar 2026

Architecture Workflow Loop

Embed a continuous architecture audit and refactor loop inside the delivery workflow to prevent structural drift during rapid experimentation.

Confidence: medium
patterns architecture-governance workflow
exploring Feb 2026

Context Injection

Collect relevant state from system stores and inject it into prompts as structured context to enable data-aware responses without fine-tuning.

Confidence: high
agentic-ai memory context
exploring Feb 2026

Deterministic Guardrail

Use a non-LLM, rule-based agent to enforce safety constraints with predictable, explainable outcomes.

Confidence: high
agentic-ai safety guardrails
exploring Feb 2026

Auto-Injection of History

When current session state is empty, automatically backfill chat context with recent stored history to preserve continuity.

Confidence: high
agentic-ai memory history
exploring Feb 2026

Graceful Fallback

Wrap LLM calls with error handling that returns safe defaults so the system stays usable when generation fails.

Confidence: high
agentic-ai safety reliability
exploring Feb 2026

Dual-Path Routing

Classify user intent into a small set of routes and dispatch to specialized handlers to optimize responses for different request types.

Confidence: high
agentic-ai planning routing
exploring Feb 2026

Heuristic-Driven Arbitration

Use a control point to adjust LLM-produced outputs based on deterministic safety signals from other components.

Confidence: high
agentic-ai control safety
exploring Feb 2026

Implicit Tool State Mutation

Accumulate complex outputs by having a tool call mutate agent state during generation, then read the final result from that state.

Confidence: high
agentic-ai execution tools
exploring Feb 2026

Orchestrator-Managed Pipeline

Use a central orchestrator to run a fixed, dependency-ordered sequence of agents so analysis steps remain consistent and traceable.

Confidence: high
agentic-ai control orchestration
exploring Feb 2026

Prompt-as-Code Separation

Store prompts outside agent code and load them at runtime to improve maintainability and enable language variants.

Confidence: high
agentic-ai contract prompts
exploring Feb 2026

Structured Output via Schema

Define a formal schema for agent outputs and validate LLM responses against it to keep results machine-readable and reliable.

Confidence: high
agentic-ai contract structured-output
exploring Feb 2026

LLM Provider Abstraction Boundary

Introduce a provider-agnostic interface so agents can switch LLM backends without rewriting orchestration logic.

Confidence: high
patterns llm abstraction
exploring Feb 2026

Prompt–Schema Contracts as First-Class Architecture

Treat prompts, schemas, and structure locks as architectural contracts that constrain outputs and reduce drift.

Confidence: high
patterns prompts schemas
exploring Feb 2026

Stabilise the Core Loop Before Adding Modalities

Defer multimodal features until the text-based agent loop and contracts are stable, to avoid compounding uncertainty.

Confidence: medium
patterns stability scope
exploring Feb 2026

Derived Documentation from Evidence

Treat docs as derived state by regenerating AS_BUILT and DELTA_BACKLOG from executable evidence to prevent drift.

Confidence: high
patterns evidence documentation

Patterns are not just ideas—they are proven solutions to recurring problems in specific contexts.