When we launched Fullstory MCP in late March, we were thrilled by the response. Thousands of builders across companies of all sizes have now joined our beta program. Teams are wiring Fullstory behavioral context directly into their development workflows, support pipelines, and AI agents. The video below walks through two of the use cases we’re seeing most.
→ MCP was one of three products we announced at our Summer Release Event. See what else is new.
Learning from our customers
The beta has surfaced a few clear patterns in how teams are putting MCP to work. See the prompts that are working for our beta users in the MCP prompt library.
Proactively detect and fix issues: Set up an automated process to monitor for broken user flows and deliver rich, contextual insights directly where your teams work. Automatically attach session context to Engineering and Support tickets to accelerate resolution.
Zac Sheffer, CEO of The Last Unicorn, did this. Every time a ticket is created in Linear, an automation pulls the associated Fullstory session, attaches it to the ticket, and generates a proposed fix, including a Playwright script built from synthetic data that mirrors the user’s actual experience. Engineers pick up tickets with the reproduction steps already done.
Release monitoring without manual digging: After a deploy, you shouldn’t have to manually trace whether a new issue was introduced. With MCP, you can ask it to check for regressions across a release window. It discovers escalating problems, pulls aggregated stats on who was affected and on which devices, and combines that with actual session data to give you a synthesized diagnostic, including a recommended priority for action.
One knowledge base, no more silos: Unify your digital experience insights so your entire team can get answers with natural language, without needing to be trained on every tool. Richard Coombes, VP of Technology at Equals Group, described using MCP alongside CloudWatch logs and Datadog to detect issues before customers report them. For a payments company where a delayed resolution can mean a missed salary or a supplier that doesn’t get paid on time, that speed matters.
Agentic Session Review
We recently announced Agentic Session Review. Normally, even if you’ve identified a trend, say a spike in checkout errors, you still have to watch sessions manually to understand what’s actually happening. Agentic Session Review removes that step. Prompt an AI agent to review a batch of sessions on your behalf. It analyzes DOM events, network calls, and user interactions, then returns a synthesized summary directly to your command line.
To avoid burning through your context window or distorting the AI’s reasoning, it uses a zoom-switch-layer model. The agent starts with a high-level session map, zooms in on specific timestamps to view actual pixels and the page’s accessibility tree, and can pull in network traffic and console logs at the exact moment an issue occurred.
Agentic Session Review is now in early access via Fullstory MCP.
Developer documentation is now live
Developer documentation is now live for all paid Fullstory customers. MCP is also open to self-serve onboarding: anyone with a paid Fullstory account can join the beta and get started today.
