Set Up Agent Studio Triggers for Real-Time AI Starts
Agent Studio Triggers are event listeners that launch a Sparkforge AI agent the instant a form is submitted, a tag changes, or a chat message arrives. Add a Start trigger to the agent canvas, configure the event, test, and publish.
Agent Studio Triggers fire AI agents on real platform events instead of polling or external webhooks. When a contact submits a form, gets tagged, or sends a chat, Sparkforge records the event and starts the agent in the background — pages and APIs stay fast while the agent responds immediately.
What triggers do
A Start trigger attaches to the agent's Start node and listens for one of three platform events. Three types are supported today:
- Form submitted
- Lead tag (added or removed)
- Chat message
When the event fires, the agent runs from its Start trigger through the rest of the flow.
Why use them
- Real-time: respond the moment a contact acts.
- No code: remove polling scripts and external webhooks.
- Consistent entry: standardize how every agent starts.
- Non-breaking: existing API-invoked agents keep working unchanged.
Before you start
- Open the correct sub-account or location.
- Confirm access to AI Agents > Agent Studio.
Set up a trigger
- Open Agent Studio. From the left sidebar click AI Agents, then the Agent Studio tab.
- Create or open an agent. Click Create Agent for a new one or pick an existing agent.
- Open the canvas. Use Add New Node or locate the start of the flow.
- Add a Start trigger. Choose Start trigger under Flow Control.
- Pick the trigger type. Form submitted, Lead tag, or Chat message.
Form submitted trigger
- Starts the agent when any selected form is submitted.
- Exposes submitted form fields as variables for later steps.
- Use the dropdown to choose forms, or click Add new to create one.
Lead tag trigger
- Starts the agent when a tag is added or removed.
- Set Tag action to Tag added or Tag removed.
- Pick the specific tag, or create one with Add new.
- For bulk tagging, run a small test segment first to avoid mass agent runs.
Chat message trigger
- Requires no extra configuration — add the trigger and publish.
- Provides message text, sender details, and conversation context.
Validate and go live
- Test: simulate the trigger to confirm variables and payload. No live contacts are affected.
- Save: store changes as a Staging version. Save after each major edit.
- Publish: push the agent live. Real events start the agent immediately. Confirm the Start trigger connects to the first node before publishing.
Frequently Asked Questions
Do triggers change how my existing API-invoked agents run?
Where do I turn triggers on?
Can I test before going live?
What data is available when a trigger fires?
What if a form or tag isn't in the dropdown?
Was this article helpful?
Thanks for the feedback.