Amplitude unveils MCP server & expands AI Agents open beta
Amplitude has announced the public release of its Model Context Protocol (MCP) server alongside an open beta for its AI Agents, allowing customers to access behavioural data within AI environments.
The introduction of MCP means that companies can now integrate Amplitude's behavioural data directly into AI environments, such as Claude, Cursor, or any compatible MCP client. This integration enables users to query data using natural language, with examples including questions like "What features drove the most retention this month?" or "What caused the drop in sign-ups last week?" Answers are generated instantly from live Amplitude data.
The company states that this development will minimise the need for users to switch between separate dashboards, enabling organisations to spend more time analysing customer behaviours and identifying opportunities for product improvement.
Behavioural context in AI
Wade Chambers, Chief Engineering Officer at Amplitude, described the importance of accurate behavioural context in digital product environments, commenting on the protocol's introduction:
"Behavioural context is one of the most critical inputs behind every great product and digital experience. The problem is, AI tools are only as good as the context they leverage," explained Wade Chambers, Chief Engineering Officer at Amplitude.
Chambers added:
"Amplitude already captures and organises behavioural data across the full user journey. So by extending this into MCP-enabled environments like Claude, we're giving AI direct, governed access to that data. This makes the context instantly available and marks a major step toward the insights-to-actions loop becoming automatic."
With MCP, AI tools can source data about the entire user lifecycle directly from Amplitude, reducing manual effort and potentially increasing the speed at which organisations can respond to user trends, challenges, and opportunities.
Customer feedback
Early adopters are claiming benefits from using the MCP server. Moss Pauly, Technical Product Manager, Data & AI at Zip, described the impact of MCP on internal analysis and data accessibility:
"Amplitude's MCP is a real game-changer," said Moss Pauly, Technical Product Manager, Data & AI at Zip. "We've been able to embed our internal domain knowledge-how certain events are used to drive complex analysis-directly into an LLM's context and have it interact seamlessly with Amplitude. It's opening up access to cross-domain insights that previously only a few specialists could uncover. It's a genuine step change in democratising data across the organisation."
According to Zip, the MCP has improved the sharing of insights within teams who previously had limited access to such analytical capabilities.
AI Agents open beta
Amplitude has also expanded access to its AI Agents, following a limited beta in June. The open beta gives all customers two AI Agents: Dashboard Agent and Session Replay Agent.
The Dashboard Agent automatically analyses dashboards, summarising trends, highlighting key metrics, and identifying areas that require attention. It can detect significant week-over-week or month-over-month changes and perform root-cause analysis on shifts in metrics. Insights are delivered proactively via Slack, email, or within Amplitude, and users can follow up with natural language queries to generate new charts or seek further clarification.
The Session Replay Agent carries out qualitative analysis of user session replays at scale. It is designed to surface behavioural patterns, identify points of friction such as rage clicks or user errors, and provide a curated playlist of relevant replays for swift investigation. Insights from this agent are also shared proactively through Slack, email, or directly in the platform.
Availability and future plans
Amplitude has confirmed that the MCP server is now available to all customers. The expanded AI Agents beta is rolling out, with further enhancements and deeper MCP integration anticipated in the coming months.
