Creating and managing a workflow in AgentSpot
Workflows are designed to automate your repeatable, time-consuming tasks.
Agents vs. workflows
Before building, it helps to understand how agents and workflows complement each other:
| Feature | Description |
|---|---|
Agents |
Think of these as people. They have personalities and specific expertise (for example, SEO, design, or sales support), and you interact with them through conversation to get specialized help. |
Workflows |
Think of these as tasks. These are automated, repeatable processes (for example, a morning briefing or deal monitor) that run in the background to handle recurring work and data synthesis. |
Workflow concepts
Every workflow in AgentSpot is built on three core concepts:
-
Trigger: Workflows run on a schedule. You decide how often the workflow runs (for example, every morning at 8:00 AM).
Event-based triggers (for example, a new ticket appearing in ServiceNow) are planned for a future release.
-
Sub-agents: A workflow is a team of specialized sub-agents, each focused on one tool or data source. Sub-agents run in parallel to process tasks as efficiently as possible.
-
Output: The final action the workflow takes, such as posting a summary to a Slack channel or sending an email.
Creating a new workflow
-
On the AgentSpot home page, scroll to the bottom and click Create new workflow.
The workflow builder opens — it works the same way as the agent builder.
-
In the prompt field, describe the task you want to automate in plain language.
Example 1. Example promptEvery morning, look at my meetings from yesterday, find the transcripts, check my emails, Slack, and Jira, and help me prioritize what to work on. Suggest draft responses for common questions. Then look at today’s meetings — for customer-facing meetings, get external context on the company and check relevant Slack channels. Send a short summary to my Slack channel and a detailed brief in Canvas.
-
Answer any clarifying questions the builder asks, such as which Slack channel to post to or how to identify a specific type of meeting.
The builder uses your answers to configure the workflow correctly. AgentSpot also applies smart defaults where it can.
-
Review the sub-agents the builder creates in the canvas.
The builder breaks your workflow into specialized sub-agents, each focused on one tool or data source. Where possible, sub-agents run in parallel to process tasks efficiently. The final sub-agent synthesizes all results into a human-readable output.
You can click any sub-agent to review or edit its instructions and connectors.
-
Set a schedule for the workflow.
Enter the time and frequency you want the workflow to run.
The scheduler currently uses UTC. You must convert your preferred local time to UTC before saving. -
Click Save.
Testing and scheduling
Before activating your workflow, run a test to verify it is working as expected.
-
Click Test run.
-
Monitor progress in the canvas. The workflow moves through each sub-agent in sequence. The left panel shows run history; the center panel shows the full workflow graph; the right panel shows details for the currently selected step.
-
Review the output. AgentSpot shows you exactly what would have been sent to your output channel.
-
If the results look correct, the workflow is active and will run on the schedule you set.
Managing existing workflows
Your workflows appear in the workflows list on the AgentSpot home page.
Click any workflow to open it. The workflow view has three panels:
-
Left panel: Run history. View past runs to verify the workflow is performing as expected.
-
Center panel: Workflow graph. Shows the full orchestration — every sub-agent and the connections between them.
-
Right panel: Step details. Click any step in the graph to see its specific inputs, outputs, and status.
Editing a workflow
Workflows are designed to be iterated on. After reviewing your results, you can refine the workflow at any time.
-
In the workflows list, find the workflow you want to update.
-
Click Edit.
The builder reopens with the full history of your previous conversation.
-
Continue chatting to describe the changes you want — for example, adjusting the output format or adding a new data source.
-
Click Save when you are satisfied with the changes.
-
Run a test to verify the updated workflow behaves as expected.
| Expect to iterate. It typically takes a few refinement cycles to get a workflow producing exactly the output you want. |