Browser Workflow Quickstart

Before You Start

• Sign in at platform.ramain.ai with Builder or Admin access.
• Ensure your organization has credits available.
• Pick a simple repeatable task, such as logging into a portal, completing a form, or extracting a table.

Step 1: Create A Workflow

1. Open Workflows home.
2. Select New Workflow and complete launch preflight if prompted.
3. The workbench opens on the Browser tab with a new executor session.

You now have an empty draft workflow tied to this authoring session.

Step 2: Author With The Executor

1. Describe the task clearly, including the starting URL, credential approach, and success criteria.
2. Watch the live browser as the agent navigates, clicks, types, and handles popups.
3. Answer confirmation or missing-input prompts from the agent.
4. Use Stop if the run goes off track, refine the instruction, and continue.
5. Repeat until the task completes successfully from end to end.

Tip: one focused goal per authoring pass compiles more cleanly than a kitchen-sink demonstration.

Step 3: Compile The Graph

1. Switch to the Editor tab.
2. Compile the latest successful trace.
3. Review the graph for semantic agent blocks, inputs, and connected edges.

If the graph looks wrong, return to Browser, correct the demonstration, and recompile.

Step 4: Save Your Draft

1. Select Save and give the workflow a descriptive name.
2. Confirm its status remains draft.

Saving preserves the graph and chat history; it does not publish the workflow.

Step 5: Publish

1. Resolve any Editor validation errors.
2. Select Publish when the graph matches the demonstrated task.
3. Confirm the status changes from draft to saved.

Publishing is the only transition from draft to saved.

Step 6: Relaunch

Return to Workflows home and launch the published workflow with fresh inputs. Use History to inspect its trace, result, and any failed step.

Common First-Run Issues