For a Java application that must survive restarts, wait for people or external events, or coordinate work over minutes or days, use Spring AI for model interaction and Dapr Workflows for durable orchestration. Define the business process as a workflow, and put model calls and external actions in workflow activities. The workflow tracks progress, waits, and retries; activities perform the work whose results the workflow can recover from its history.
This is a composition of documented capabilities, not a turnkey integration recipe: the Spring AI and Dapr materials cited here do not present an official combined reference application.
What Spring AI and Dapr each do
These projects address different layers, so using both does not mean one replaces the other.
Spring AI: model interaction
Spring AI provides ChatClient, a fluent API for communicating with AI models, with synchronous and streaming programming models. Its reference also discusses agentic workflow patterns and recommends using the simplest pattern that meets the requirements. In a durable design, Spring AI is the model-facing part of an activity—not the process engine responsible for keeping a multi-step job alive.
#1 Best Overall
Dapr Workflows: durable process state
Dapr Workflows provides workflow and activity definitions, timers, scheduled tasks, waits for external events, retries, and recovery behavior. The Dapr Spring Boot integration registers workflow and activity beans and exposes DaprWorkflowClient for scheduling workflow instances and raising events. Dapr’s guide, How to: Author and manage Dapr Workflow with Spring Boot, describes that integration surface.
How to structure a long-running agent
Represent the task as explicit business steps rather than one large prompt-and-tool loop. A typical process might accept a request, analyze it, perform a tool action, validate the result, and pause for approval or a callback when needed.
- Define the process and its state. Decide what the workflow must remember between steps: request identifiers, completed outcomes, approval status, and other data needed to resume. Keep workflow inputs and outputs serializable and stable.
- Put nondeterministic work in activities. Invoke Spring AI from an activity. Put network calls, database writes, and other external side effects in activities too. Workflow functions can be replayed to rebuild local state; completed activity results can be recovered from workflow history. Direct model calls or side effects in replayed workflow code could otherwise happen again.
- Make repeatable actions safe. A workflow may resume or retry work. Make activities idempotent where possible, or use a task execution key or another deduplication mechanism supported by the SDK API you select. Dapr’s Java workflow documentation describes task execution keys as useful for idempotency and state management.
- Model waits explicitly. Use durable timers for scheduled delays and external events for callbacks or human decisions. Dapr documents that event signals can be retained in workflow history until the workflow reaches its wait, so an event need not arrive at precisely the same moment as the wait.
- Choose retry behavior deliberately. Workflow retry policies retain retry state across application restarts. Dapr Resiliency policies address a different layer and are not themselves durable across application restarts. Decide which failures should retry at the activity or workflow level, and avoid combining retry layers in a way that repeats an external action unexpectedly.
- Provide a way to inspect and resume work. Schedule the workflow and raise events through
DaprWorkflowClient. Expose an application-appropriate status path so callers can see progress and, where relevant, submit a callback or approval. Dapr examples illustrate querying workflow instances, but the endpoint shape depends on the application.
Why replay changes where code belongs
A Dapr workflow can be unloaded and later replayed to reconstruct its local variables. During replay, completed tasks are satisfied from recorded history. That is different from ordinary request handling, where developers often assume a function body runs just once per request.
Model output is not guaranteed to be identical across calls, and an external tool may commit an irreversible action. Keeping those operations behind activity boundaries allows the workflow to continue from recorded outcomes rather than casually issuing them again. This is a design consequence of Dapr’s documented replay and activity model; it is not a claim about a tested Spring AI–Dapr sample.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Rank #3
When a workflow-backed design is worth the added machinery
Choose workflow orchestration when a task must persist across process restarts, wait for a person or external event, coordinate multiple services, or provide inspectable progress and recovery. For a short, synchronous prompt-and-response interaction, a regular Spring AI request path is usually the simpler fit.
| Consideration | Synchronous Spring AI request | Spring AI activity in a Dapr workflow |
|---|---|---|
| Typical duration | One request and response | A process that may span minutes or days |
| Recovery | Handled by application-level retry or restart logic | Can resume using durable workflow history |
| Waiting | Request remains part of the immediate interaction | Can use durable timers, callbacks, and approval waits |
| Operational surface | A comparatively simple request service | Workflow state and history plus the runtime components required by the deployment |
| Control and complexity | Less orchestration setup; process control remains in application logic | Explicit workflow progress and retry behavior, with additional orchestration complexity |
This is a comparison of documented capabilities, not a performance benchmark. For consequential actions, a human approval step can be modeled as a workflow event; Dapr Agents documentation illustrates human-in-the-loop pauses, while the same orchestration concept can be implemented with Dapr Workflows in a Spring application.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Version compatibility and project boundaries
The Dapr Spring Boot workflow guide labels its integration alpha and says it requires Spring Boot 3.x or later. The Java SDK repository records compatibility changes across SDK lines, including a Spring Boot 4 target for a later line and guidance for Spring Boot 3.5 users. Check the current compatibility matrix and align the SDK, Spring Boot, and dependency-management versions rather than copying a version from an older tutorial.
Dapr Agents’ DurableAgent is a separate framework. Its surfaced implementation examples are Python-based; they can illustrate durable-agent patterns but do not establish that Spring AI uses DurableAgent or that it is the Java integration described here. The documented Spring Boot workflow integration is the relevant Dapr surface for this architecture.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteQuick Recap
Best Value
- Book - powershell for sysadmins: workflow automation made easy
- Language: english
- Binding: paperback
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
