Embrace MCP Server now in beta: Bring real-user observability into your AI workflows

The Embrace MCP Server implements the Model Context Protocol (MCP), a standard for securely sharing rich, structured context between tools and AI agents.

In practice, this means production-grade observability signals from Embrace can now be queried and consumed by AI systems outside the Embrace dashboard.

Instead of observability living in one place and AI tools living in another, MCP bridges the two. 

Best of all, the data is grounded in real user behavior and experiences, which is what makes Embrace unique as a frontend observability solution. 

Observability data is uniquely powerful, but historically hard to use outside of proprietary platforms and dashboards. At the same time, AI tools are only as useful as the context they can access. Something is needed to bridge the gap here, linking the power of observability data with the capabilities of AI tools to automate, streamline, and improve engineering workflows. 

MCP solves this by standardizing how context flows between systems, much like the way that OpenTelemetry standardized how telemetry is collected.

With MCP, Embrace observability data becomes:

• Portable: usable across AI agents, copilots, and internal tools
• Structured : pre-processed signals, not raw, overwhelming telemetry
• Interoperable: no lock-in to a single workflow or platform

This makes observability usable where decisions actually happen, whether that’s in your IDE, BI dashboard, or other internal, AI-powered tool. 

With the MCP Server enabled, teams can begin to power AI-driven workflows focused on their application health, such as:

• Asking an AI assistant about trending crashes, slow sessions, or user-impacting issues
• Correlating frontend issues with recent releases or code changes automatically
• Feeding real-user performance signals into internal tools, notebooks, or dashboards
• Enabling agents to perform repetitive observability checks and surface prioritized insights

The heavy lifting – aggregation, signal extraction, and context preparation – happens before the data reaches an AI agent, so models can focus on synthesis and guidance rather than raw analysis.

We didn’t build MCP to ship “another AI feature.” We built it as a long-term foundation.

Different teams will adopt AI differently, some with custom agents, others with third-party tools, many with a mix of both. The Embrace MCP Server is designed to meet teams where they are, integrating cleanly into existing workflows rather than forcing a new one.

The Embrace MCP Server is available in beta now. You can learn how to configure it, explore supported use cases, and start integrating Embrace observability data into your AI workflows through our docs. 

And if you want a deeper look at why we built MCP, and our broader philosophy on AI and observability, check out this blog from our CTO, Fredric, and AI product leader, Kasey.