Explore topics

Topic

Durable AI Agents

A chat completion forgets everything the moment the request ends. Durable agents don't: they checkpoint progress between steps, park while they wait for human approval, wake on a schedule, and keep memory across sessions. This topic pairs two runtimes — Workflow DevKit patterns for durable step execution (sequential, parallel, evaluator loops, orchestrator-worker) and Eve agents for the full durable lifecycle: tenant-scoped memory, cron-driven digests, and approval-gated operations that resume safely. Start with the simple Eve agent, then add the capability your system is missing.

6 patterns4 featured
Open featured previewSee pricing

Featured patterns

A curated starting point for this topic — open a live preview, then adapt the source.

4 curated
Orchestrator-Worker Workflow Pattern preview
01Intermediate

Orchestrator-Worker Workflow Pattern

Demonstrates an orchestrator-worker workflow pattern using Workflow DevKit. Features parallel execution of specialized workers for different change types (create, modify, delete).

  • Coordinate create, modify, and delete workers from an orchestrator
  • Run specialized workers in parallel under Workflow DevKit
  • Plan multi-file change sets before worker execution
generateObjectaiworkflow
Open live preview
Sequential Workflow Pattern preview
02Intermediate

Sequential Workflow Pattern

Demonstrates a sequential workflow pattern using Workflow DevKit. Features step-by-step execution with quality gates and conditional regeneration.

  • Execute AI steps sequentially with Workflow DevKit
  • Insert quality gates between generation stages
  • Conditionally regenerate content when quality checks fail
generateObjectgenerateTextaiworkflow
Open live preview
Evaluator Workflow Pattern preview
03Intermediate

Evaluator Workflow Pattern

Demonstrates an evaluator-optimizer workflow pattern using Workflow DevKit. Features iterative improvement through AI evaluation and optimization loops.

  • Run an evaluator-optimizer loop inside Workflow DevKit
  • Score outputs with structured evaluation schemas
  • Iterate translation or content quality until thresholds pass
generateObjectgenerateTextaiworkflow
Open live preview
Workflow - URL Analysis preview
04Beginner

Workflow - URL Analysis

Analyze URLs through AI prompt chaining. Scrapes content, analyzes it, and generates summaries.

  • Chain scrape, analyze, and summarize steps for a URL
  • Stream per-step workflow status with typed data parts
  • Produce structured analysis via generateObject before summarization
streamTextconvertToModelMessagescreateUIMessageStreamai
Open live preview

More patterns

Additional patterns that fit this topic.

2 more
01
Parallel Review Workflow PatternDemonstrates a parallel review workflow pattern using Workflow DevKit. Features concurrent code reviews (security, performance, maintainability) with result aggregation.
02
Routing Workflow PatternDemonstrates a routing workflow pattern using Workflow DevKit. Features query classification and conditional routing to specialized handlers based on type and complexity.

Related topics

Continue with adjacent topics that share the same building blocks.

AI Agent Orchestration

7

Working orchestration code, not diagrams: routing, sub-agent delegation, parallel fan-out, and orchestrator-worker patterns in TypeScript on the AI SDK. Live previews + full source.

Explore topic

AI Agent Tools

22

Give agents real capabilities: tool() definitions with lifecycle hooks, dynamic tools, tool-call repair, web search (Exa, Firecrawl), scraping, PDF ingestion, and human approval gates.

Explore topic

AI SDK Examples

52

Complete, running examples for the AI SDK: generateText, streamText, structured output, tool calling, agent loop control, prepareStep, and tool-call repair. Copy the source, keep shipping.

Explore topic

Browse by category

Jump into the catalog filtered by category.

Agent PatternsWorkflows

You’ve reached the end of Durable AI Agents.