- Get started
- Build
- Deploy and run
- Examples
- Reference
FAQ
Answers to common questions about Maestro Automate, each with a link to the page that covers the topic.
Getting started
What is Maestro Automate?
Maestro Automate is a product for building workflows in which AI agents and APIs work together. What is Maestro Automate? describes it.
Where can a workflow be built?
In Studio Web, in your browser, with nothing to install. VS Code and Studio Desktop are also available. Where to build compares the three.
Do I need to install anything?
No. Studio Web runs in the browser. Where to build describes the environments that run on your computer, if you prefer one.
Can a coding agent build the workflow for me?
Yes. In VS Code, a coding agent such as Claude Code, Codex, Cursor, or Copilot can write the workflow file. In Studio Web and Studio Desktop, Autopilot (Preview) can build the workflow from a description. Build in VS Code describes the VS Code setup.
What is the fastest way to try it?
Your first workflow builds and debugs the order intake workflow in about 15 minutes.
What does a workflow need before it can run?
At least one trigger. Triggers describes the trigger types.
Running and cost
How is Maestro Automate priced?
Per run. Building, testing, and evaluating are free, and every plan includes a number of runs each month. Pricing lists what each plan includes.
How long can a run last, and how long is its data kept?
A run can last up to 24 hours, and it's kept for 30 days after it ends. The agents and other work it calls, their traces, and audit events have their own retention settings, the same as in Maestro Orchestrate. How long run data is kept describes them.
Can a run be retried or canceled?
Yes. A run that failed at a step can be retried from that step, and any run can be canceled. Retry and cancel describes how.
How many runs can be in progress at the same time?
Up to 50. Limits and fair use lists the limits.
Building and testing
How is a workflow tested before it is deployed?
In debug mode. Select Run Debug (▶) in the bottom toolbar, and the execution trace shows what every step received and returned. Debugging basics describes how to read a run.
Can the output of an AI agent be tested automatically?
Yes, with evaluations. Evaluations describes datasets, evaluators, and runs.
How is a human's approval added to a workflow?
With a Quick Form node, from the Human category. Add human approval to a workflow builds a complete example.
Can a workflow use my existing robots, agents, and API workflows?
Yes, as nodes. A workflow can call RPA workflows, agents, API workflows, and Business Process Model and Notation (BPMN) processes. Call existing work lists them.
Can I keep personal data out of an agent's prompts?
Yes, with guardrails on the agent node. Autonomous Agent describes them.
What happens when a step fails?
In a deployed workflow, the run pauses at the failed step and waits for you to retry or cancel it, as Retry and cancel describes. To let the workflow handle the failure on its own, connect the node's error path to other steps. Error handling describes how.
Moving to Maestro Orchestrate
Should a new process start in Maestro Automate or Maestro Orchestrate?
Start in Maestro Automate when each run finishes within 24 hours. Start in Maestro Orchestrate when the process waits days for a human or an event, someone must act on runs in progress, or it needs the BPMN or Case canvas. Maestro Orchestrate or Maestro Automate? compares the two products.
Can a workflow move to Maestro Orchestrate?
Yes, and converting is optional. The workflow isn't rebuilt, and you can convert it back. Convert to Maestro Orchestrate describes when to convert and what changes.
Does Maestro Automate keep fewer audit logs than Maestro Orchestrate?
No. Audit events are captured the same way in both products. The difference is the run itself, which Maestro Automate deletes 30 days after it ends. How long run data is kept describes the retention settings for each part of a run's data.
Next steps
- Troubleshooting: fixes for the most common errors.
- Community: ask other builders when your question isn't here.
- Getting started
- What is Maestro Automate?
- Where can a workflow be built?
- Do I need to install anything?
- Can a coding agent build the workflow for me?
- What is the fastest way to try it?
- What does a workflow need before it can run?
- Running and cost
- How is Maestro Automate priced?
- How long can a run last, and how long is its data kept?
- Can a run be retried or canceled?
- How many runs can be in progress at the same time?
- Building and testing
- How is a workflow tested before it is deployed?
- Can the output of an AI agent be tested automatically?
- How is a human's approval added to a workflow?
- Can a workflow use my existing robots, agents, and API workflows?
- Can I keep personal data out of an agent's prompts?
- What happens when a step fails?
- Moving to Maestro Orchestrate
- Should a new process start in Maestro Automate or Maestro Orchestrate?
- Can a workflow move to Maestro Orchestrate?
- Does Maestro Automate keep fewer audit logs than Maestro Orchestrate?
- Next steps