> ## 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 1: Connect live user activity > Help the onboarding agent understand what a user has and has not tried so it can guide them to the right next step. Before Autoplay can offer useful guidance, it needs to understand what the user has and has not tried inside your product. That context informs what to suggest next and prevents the support agent from asking users to repeat steps they already completed. Your session replay provider already captures this activity. Autoplay connects to that existing source and turns recent page visits, clicks, and outcomes into a clearer picture of the user's current journey. Autoplay currently supports [PostHog](/recipes/posthog/how-to-setup) and [Amplitude](/recipes/amplitude/how-to-setup), with support for [FullStory](/recipes/fullstory/how-to-setup), [Datadog](/recipes/datadog/how-to-setup), and [Pendo](/recipes/pendo/how-to-setup) coming soon. You keep the provider already installed in your product. The integration simply lets Autoplay use that activity. ## From isolated actions to understandable behavior A single click rarely explains what someone is trying to accomplish. A sequence of actions is more useful: 1. The user opens the integration settings. 2. They select a data provider. 3. They return to the same setup page several times. 4. No successful connection is detected. Together, those actions suggest that the user is trying to connect an integration and may be having trouble completing it. Session replay captures recent activity. Autoplay MCP understands what the user has and has not tried, turning the existing support agent into an onboarding agent. Autoplay keeps this recent activity for a limited period so it can reason about the current journey without creating a permanent archive of the customer's session replay data. ## Give your support agent live product context Once the user opens your existing support agent, Autoplay MCP gives it an understanding of what the user has tried, what they have not tried, and where they are in the workflow. The conversation can begin with the right next step instead of asking the user to start from the beginning. Autoplay MCP is the brain that makes the existing support agent an onboarding agent. It gives the agent a consistent way to ask questions such as: * What has this user done recently? * Which onboarding workflow are they completing? * What have they already finished? * What is the most relevant next step? * Is there a visual tour that can guide them through it? The support agent does not need a custom integration for every type of activity or workflow. Through Autoplay MCP, it can use the understanding already built from live behavior and workflow progress to decide how to guide the user next. The two connections serve different purposes: | Connection | Purpose | | -------------------------- | -------------------------------------------------------------------------------------------------------- | | Session replay integration | Sends recent product activity to Autoplay | | Autoplay MCP | Gives the support agent access to recent activity, workflow progress, next steps, and available guidance | This means the support agent can respond to “Why isn't my integration working?” with awareness of the setup steps the user has already taken. It can explain the relevant next action or launch visual guidance without asking the user to reconstruct their journey. ## Why this matters Without live activity, even a capable support agent has to rely on what the user remembers to describe. With Autoplay, it can understand the immediate situation and provide help that matches the user's actual journey. At this stage, Autoplay understands what happened. The next step gives that behavior meaning by defining what a successful journey should look like. Define what different users should learn and what successful progress looks like.