Think inside your AI world.
How do I connect Amplitude to Claude?
Amplitude's MCP server hands Claude the charts, dashboards, experiments, flags and metrics — the read layer of a mature analytics practice.
Direct answer: connect Amplitude's MCP server from Claude's Settings → Connectors and authorise ('claude mcp add' in Claude Code). Claude then reads charts, dashboards, experiments, feature flags and metrics from your project — the questions you would normally answer by assembling views, answered in conversation.
Metrics without the dashboard shuffle
How the chart moved since the release, what the experiment's arms show, which flags are live in production, whether the dashboard's north-star panel actually shifted — read from the project directly. Analysts save clicks; founders skip the interface entirely.
One metric, two honest definitions
Worked case: your team ratified a retention definition — day-30 return, not day-7, is the number decisions cite, because day-7 once flattered a cohort that had evaporated by the month mark. Amplitude computes both flawlessly. Which one is decision-grade in your company is a ruling about the metrics, invisible to the metrics.
Charts read under your definition
Through Unl the retention read arrives pre-framed: day-30 for the March cohort, the figure your ruling makes citable — with day-7 labelled as the flattering cousin your team retired for decisions. Metric disputes end at read time rather than mid-meeting.
Amplitude to Claude reads the metrics; Amplitude through Unl reads the metrics under the definitions you ratified — so the number that arrives is the one your decisions are allowed to cite.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. The reach lane is live: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What is on Amplitude's MCP read surface?
Charts, dashboards, experiments, feature flags and metrics — the observational layer of the product-analytics stack, readable in conversation.
Does the server work beyond Claude?
Yes; 'amplitude mcp' connects identically from ChatGPT or any MCP client. The protocol is shared, the steps near-identical.
Why ratify a metric definition rather than document it?
Documentation waits to be consulted; a ratified definition arrives at read time, applied. When the wrong cousin of the metric shows up in a conversation, the ruling — and the cohort story behind it — shows up with it.
What if two teams genuinely need different definitions of the same metric?
Then ratify two, each scoped to where it applies, rather than forcing one definition to cover both. A single definition stretched over genuinely different questions is how a number ends up argued about in every meeting it appears in.
What this is
Think inside your AI world — you stay in command
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. A new unit of exchange between you and your AI: the Settled Why with standing that travels. Unprompted.
Your whole AI world. What you decided at the epicentre. It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector — quick to connect, in a couple of steps.
MCP native·Human settled·Model agnostic·Your data
Measured Context
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free at launch.