Plain-English report builder for Amplitude

A product manager or growth marketer who needs to answer 'how many users completed onboarding this week vs. last week, broken down by acquisition channel' knows exactly what they want to know. What they cannot do is translate that question into Amplitude's event taxonomy, funnel configuration, and chart type — especially when, as users describe, 'it is easy to misuse the product' and 'configuring accurate tracking often demands technical know-how.'

The gap exists because Amplitude's query model is built around event schemas that have to be designed upfront by engineers. The tool is sold to product and engineering leadership, but the daily users are PMs and marketers who had no input into how events were named or structured. There's no layer that bridges the user's natural language question and the underlying event data model specific to their company's implementation.

Generic AI query tools don't solve this because they don't know your company's specific event taxonomy — they can't know that your company calls a signup 'user_registered' or that 'checkout_initiated' has a property called 'plan_tier'. What's needed is a tool that ingests a company's specific Amplitude event schema and lets non-technical users build queries by describing what they want in plain language, with the output validated against actual tracked events.

This recurs every time a new PM joins, every time a team expands analytics access, and every time someone needs to answer a question quickly but doesn't have time — as one user put it — to 'understand and get fluent with the software' under time pressure. The concrete cost is decisions made on gut feel because the person who can run the query is always busy.

What to build

Build a web app that connects to a company's Amplitude account via API, ingests their full event schema and property list, and lets non-technical users describe a question in plain English, then generates and runs the correct Amplitude query, showing the result with a one-sentence interpretation.

Where to start

Target companies that use a specific Amplitude template or starter schema (e.g. the standard e-commerce tracking plan) where events are predictable enough that the translation layer works reliably out of the box before you need to handle custom schemas.

The hard part

Event schemas across companies are messy — poorly named events, inconsistent properties, undocumented conventions — so natural language to query translation will fail on edge cases and erode trust fast if the first few answers are wrong.

How it makes money

Per-seat monthly subscription for non-technical users (PMs, marketers) at ~$30–$60/seat/month, with the engineering team on a free read-only tier so they can verify query accuracy — creating an internal champion who drives adoption.