ai-examples

<!-- markdown-link-check-disable --> [](https://docs.restate.dev) [](https://discord.gg/skW3AZ6uGd) [](https://join.slack.com/t/restatecommunity/shared_invite/zt-2v9gl005c-WBpr167o5XJZI1l7HWKImA) [](https://x.com/intent/follow?screen_name=restatedev) <!-- markdown-link-check-enable --> # Examples for AI workflows and Durable Agents This repo contains a set of runnable examples of AI workflows and agents, using **Durable Execution and Orchestration** via [Restate](https://restate.dev/) ([Github](https://github.com/restatedev/restate)) The goal is to show how you can easily add production-grade _resilience_, _state persistence_, _retries_, _suspend/resume_, _human-in-the-loop_, and _observability_ to agentic workflows. So you can ship agents that stay alive and consistent without sprinkling retry-code everywhere and without building heavyweight infra yourself. The Restate approach works **independent of specific SDKs** but **integrates easily with popular SDKs**, like the [Vercel AI SDK](https://ai-sdk.dev/) or the [OpenAI Agent SDK](https://openai.github.io/openai-agents-python/). You can also use without and Agent SDK _(roll your own loop)_ or for more traditional workflows. ## Why Restate? 📄 For a gentle intro, read [the blog post "Durable Agents - Fault Tolerance across Frameworks and without Handcuffs"](https://restate.dev/blog/durable-ai-loops-fault-tolerance-across-frameworks-and-without-handcuffs/) | Use Case | What it solves | |------------------------------------|---------------------------------------------------------------------------------------------| | **Durable Execution** | Crash-safe LLM/tool calls & idempotent retries—agents resume at the last successful step. | | **Detailed Observability** | Auto-captured trace of every step, retry, and message for easy debugging and auditing. | | **Human-in-the-loop & long waits** | Suspend while waiting for user approval or slow jobs; pay for compute, not wall-clock time. | | **Stateful sessions / memory** | Virtual Objects keep multi-turn conversations and other state isolated and consistent. | | **Multi-agent orchestration** | Reliable RPC, queuing, and scheduling between agents running in separate processes. | <img src="/doc/img/patterns/parallel_tools.png" alt="Restate UI - trace of agent with parallel tools" width="900px"/> <br/> <caption><em>Restate UI showing an agent execution with parallel tool calls</em></caption> ## Full Example Catalog ### Agent SDK Integrations | Integration | Example | Description | Code | Docs | |---------------------------|---------|-------------|----------------------------------------------------------------------------------------------------------------------|------------------------------------------------------| | **Vercel AI SDK** | **Template** | A minimal example of how to use Restate with the Vercel AI SDK | [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](vercel-ai/template) | [📖](https://docs.restate.dev/ai-quickstart) | | | **Tour of Agents** | A step-by-step tutorial showing how to build resilient agents | [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](vercel-ai/tour-of-agents) | [📖](https://docs.restate.dev/tour/vercel-ai-agents) | | | **Examples** | More advanced examples that can be deployed as a Next.js app on Vercel | [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](vercel-ai/examples) | - | | **OpenAI Agents SDK** | **Template** | A minimal example of how to use Restate with the OpenAI Agents SDK | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](openai-agents/template) | [📖](https://docs.restate.dev/ai-quickstart) | | | **Tour of Agents** | A step-by-step tutorial showing how to build resilient agents | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](openai-agents/tour-of-agents) | [📖](https://docs.restate.dev/tour/openai-agents) | | **Google ADK** | **Tour of Agents** | A step-by-step tutorial showing how to build resilient agents | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](google-adk/tour-of-agents) | [📖](https://docs.restate.dev/tour/google-adk) | ### Composable AI Patterns | Pattern | Description | Code | Docs | |------------------------|-------------|------|------| | **Chaining LLM calls** | Build fault-tolerant processing pipelines where each step transforms the previous step's output | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/chaining.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/chaining.ts) | [📖](https://docs.restate.dev/ai/patterns/prompt-chaining) | | **Tool routing** | Automatically route requests to tools based on LLM outputs | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/routing_to_tool.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/routing-to-tools.ts) | [📖](https://docs.restate.dev/ai/patterns/tools) | | **Parallel tool execution** | Execute multiple tools in parallel with durable results that persist across failures | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/parallel_tools.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/parallel-tools.ts) | [📖](https://docs.restate.dev/ai/patterns/parallelization) | | **Multi-agent routing** | Route requests to specialized agents based on LLM outputs | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/routing_to_agent.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/routing-to-agent.ts) | [📖](https://docs.restate.dev/ai/patterns/multi-agent) | | **Remote agent routing** | Deploy/scale agents separately and route requests with resilient communication | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/routing_to_remote_agent.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/routing-to-remote-agent.ts) | [📖](https://docs.restate.dev/ai/patterns/multi-agent) | | **Parallel agent processing** | Run multiple, specialized agents in parallel and aggregate their results | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/parallel_agents.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/parallel-agents.ts) | [📖](https://docs.restate.dev/ai/patterns/parallelization) | | **Racing agents** | Race multiple agents against each other and use the fastest response | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/racing_agents.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/racing-agents.ts) | [📖](https://docs.restate.dev/ai/patterns/competitive-racing) | | **Human-in-the-loop pattern** | Implement resilient human approval steps that suspend execution until feedback is received | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/human_in_the_loop.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/human-in-the-loop.ts) | [📖](https://docs.restate.dev/ai/patterns/human-in-the-loop) | | **Chat sessions** | Long-lived, stateful chat sessions that maintain conversation state across multiple requests | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/chat.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/chat.ts) | [📖](https://docs.restate.dev/ai/patterns/sessions-and-chat) | | **Orchestrator-worker pattern** | Break down complex tasks into specialized subtasks and execute them in parallel | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/orchestrator_workers.py) | - | | **Evaluator-optimizer pattern** | Generate → Evaluate → Improve loop until quality criteria are met | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](python-patterns/app/evaluator_optimizer.py) [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](typescript-patterns/src/evaluator-optimizer.ts) | - | ### Other Examples | Example | Description | Code | |------------------------|-------------|------| | **MCP** | Using Restate for exposing tools and resilient orchestration of tool calls | [<img src="https://skillicons.dev/icons?i=ts&theme=light" width="20" height="20">](mcp) | | **A2A** | Implement Google's Agent-to-Agent protocol with Restate as resilient, scalable task orchestrator | [<img src="https://skillicons.dev/icons?i=python&theme=light" width="20" height="20">](a2a) | Restate currently supports 6 languages: [](https://docs.restate.dev/develop/ts/overview) [](https://docs.restate.dev/develop/python/overview) [](https://docs.restate.dev/develop/java/overview) [](https://docs.restate.dev/develop/java/overview) [](https://docs.restate.dev/develop/go/overview) [](https://docs.rs/restate-sdk/latest/restate_sdk/) The examples can be translated to any of the supported languages. Join our [Discord](https://discord.gg/skW3AZ6uGd)/[Slack](https://join.slack.com/t/restatecommunity/shared_invite/zt-2v9gl005c-WBpr167o5XJZI1l7HWKImA) to get help with translating an examples to your language of choice. ## Learn more - [Documentation](https://docs.restate.dev/ai) - [Examples on workflows, microservice orchestration, async tasks, event processing](https://github.com/restatedev/examples) - [Restate Cloud](https://restate.dev/cloud/) - [Discord](https://discord.gg/skW3AZ6uGd) / [Slack](https://join.slack.com/t/restatecommunity/shared_invite/zt-2v9gl005c-WBpr167o5XJZI1l7HWKImA) ## Acknowledgements - The DIY patterns are largely based on Anthropic's [agents cookbook](https://github.com/anthropics/anthropic-cookbook/tree/main/patterns/agents). - Some of the A2A examples in this repo are based on the examples included in the [Google A2A repo](https://github.com/google/A2A/tree/main).