State before nodes
Define the business event and source of truth before the workflow graph becomes the accidental system design.
D2 Automation Insights
Production automation starts with system design — not workflow nodes.
A practical knowledge base for deciding what to automate, defining source-of-truth and orchestration boundaries, making APIs and webhooks retry-safe, operating n8n in production and engineering evidence-grounded AI workflows.
Direct answer
How should a production automation system be designed?
Start with the business event and authoritative state. Keep deterministic rules inspectable, protect duplicate-sensitive side effects, classify failures before retrying, preserve durable recovery context and verify the final business outcome. Choose n8n, custom code, queues and AI only after those responsibilities are clear.
Start here by problem
Do not read the library in publication order. Start with the decision you need to make.
Knowledge map
Ten guides organized into four production automation tracks.
The tracks follow the system lifecycle: choose the process and boundary, make integrations safe, operate the runtime, then add AI where evidence and fallback can be controlled.
TRACK 01
Decide what to automate and where the boundary belongs
Start with the business process, authoritative state and system responsibility before choosing nodes, services or AI.
01.101.201.3
Automation Prioritization
Prioritize by business value, process stability, data readiness, ownership, failure controllability and measurable outcome.
Source-of-Truth-First Automation
Define the business event, authoritative state, deterministic rules, side effects and recovery model before implementation.
n8n vs Custom Code
Choose orchestration, custom-code and hybrid boundaries from system responsibility rather than tool preference.
TRACK 02
Make API and event-driven integrations safe
Protect data completeness and irreversible side effects across API contracts, duplicate delivery, retries and ambiguous failures.
02.102.202.3
Production API Integration
Review contracts, authentication, pagination, rate limits, validation, timeouts, idempotency and recovery before release.
Webhook Idempotency
Protect logical business events and downstream side effects against duplicate delivery, retries and ambiguous outcomes.
Retry, Backoff & Dead-Letter
Classify failures, bound retries, apply backoff and preserve terminal context for controlled replay.
TRACK 03
Operate n8n as a production system
Move from workflow execution to observable, recoverable operations with explicit infrastructure and release boundaries.
03.103.203.3
n8n Monitoring & Observability
Trace event intake, execution, dependencies, retries, queue health and the final business acknowledgement.
n8n Queue Mode
Understand control, webhook ingress, Redis coordination, worker execution, durable state and scaling boundaries.
Production n8n Readiness
Use release gates across ownership, validation, idempotency, retries, concurrency, observability, deployment and recovery.
TRACK 04
Engineer reliable AI knowledge workflows
Keep retrieval evidence, grounding, evaluation and fallback separate from model fluency when AI enters the workflow.
D2 operating principles
Five principles connect every guide in this automation cluster.
Protect side effects
Assume duplicate delivery, retries and ambiguous failures can happen; make external actions safe to repeat or reconcile.
Retry by semantics
Transient, terminal and business-rule failures should not share one retry policy.
Green run is not delivery
Execution success is a technical signal. Verify the downstream business outcome the workflow exists to create.
AI stays bounded
Use AI for interpretation where useful, while validation, durable state, side effects and recovery remain explicit.
From knowledge to implementation
Use the insights to make the decision, then move into services, proof and technology.
Automation services
Move from architecture questions into implementation, migration, integration and operating support.
ExploreAutomation case studies
See workflow-backed examples across infrastructure, integrations, AI systems, reporting and operations.
Exploren8n production stack
See the technology boundary behind production n8n infrastructure and operating controls.
ExploreAutomation architecture review
Need the knowledge translated into a production automation plan?
D2 can map process priority, source-of-truth, orchestration boundaries, APIs, reliability controls and recovery ownership before implementation scope is fixed.
Latest supporting Automation insights
New supporting articles published into this knowledge graph.
This topic cluster remains repository-owned. The articles below come from PostgreSQL, are server-rendered, and appear only after they are published and indexable.
Answer & evidence
What does D2's Automation knowledge base cover?
D2's Automation knowledge base follows the production path from business event and source of truth through API/webhook handling, idempotency, retry and recovery, queue/runtime design, monitoring, AI boundaries and production reliability. The objective is durable operating systems, not workflow count.