> ## Documentation Index
> Fetch the complete documentation index at: https://developers.autoplay.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Step 2 — Define proactive triggers

> Proactively message users in Crisp based on what they're doing in your product — coming soon.

<Info>
  **Coming soon.** This step is currently in progress. Check back shortly.
</Info>

Once you've connected real-time events in [Step 1](./step-1-connect-real-time-events), the next layer is proactive triggers — letting your Crisp support AI agent reach out to users **before** they ask, based on what they're doing in your product.

## What this step will cover

* Defining trigger conditions on your `ActionsPayload` stream (e.g. user stuck on the same workflow step, first visit to a key feature, repeated failure signals)
* Routing a firing trigger to Crisp as an outbound operator message via the Crisp REST API
* Gating triggers with the Autoplay FSM (`can_show_proactive_with_reason`) to avoid noise and respect cooldowns
* Using Crisp's conversation initiation API to open a new thread proactively when needed

## Why proactive triggers matter here

With only Step 1, your Crisp support AI agent is context-aware but still **reactive** — it waits for a visitor to send a message before it pulls their activity. Proactive triggers let the bot send the first message at exactly the right moment: when the user is visibly stuck, exploring a new feature for the first time, or resuming an incomplete workflow — without waiting for them to ask.

***

**Back to overview:** [Crisp AI tutorial](./)