Revenue by month
Paid orders, last 12 months, EUR.
Early access · validating the idea
Self-service BI for teams without a data team. We're validating an AI agent that plugs into any database you already run, read-only, learns what every table means, asks when something is unclear, and turns plain questions into dashboards anyone can build. If enough teams want it, we'll build it.
Paid orders, last 12 months, EUR.
orders.status holds C, X and P. Your “Monthly sales” report only counts C. Should revenue include only completed orders?
This quarter, completed orders.
From your schema and 14 imported reports.
We're validating the idea. Enough signups and we build it, and you get first access. No spam.
The problem
Most companies already have a perfectly good database. What they don't have is the time or the people to model it, document it, and answer every “can you pull this for me?” that lands in the queue.
Every question waits for the one person who knows SQL. By the time the answer arrives, the decision has already been made on gut feel.
Semantic layers, data models, a consultant. It's a big project before anyone sees their first chart, and it goes stale as the schema changes.
flag_2 meansCryptic column names, half-documented joins and duplicate tables make generic AI chat tools confidently wrong. Guessing isn't good enough for numbers you'll act on.
How it would work
Point it at Postgres, SQL Server, Snowflake or any other database. It's designed to be strictly read-only: it never writes, and there's nothing to migrate.
Hover, tap, or focus to exploreIt reads every table and column, works out which data can answer which questions, and reverse-engineers joins from your existing reports. By design, only schema metadata leaves your environment.
Hover, tap, or focus to exploreTwo columns that could both be “revenue”? A status code nobody documented? You get a short, specific question, and the answer is remembered for everyone.
Hover, tap, or focus to exploreType a question, get a chart, and drag it onto a dashboard. Anyone on the team can do it, with the SQL one click away for those who want to check it.
Hover, tap, or focus to exploreWho it's for
See revenue, retention and margins yourself, without waiting on a hire you haven't made yet.
Stop exporting to spreadsheets every Monday. Ask the question once and pin the answer to a live dashboard.
Hand the ad-hoc requests to the agent and spend your week on the work only you can do.
Give customer success and product a safe, read-only way to explore the production replica.
The goal is any database: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, Snowflake, BigQuery, Redshift, Databricks, ClickHouse and more. If your team can query it today, the agent should be able to map it. Tell us what you run when you sign up, as it shapes what we support first.
No. That's the point. The agent builds the map of tables, metrics and relationships itself, using your schema and the reports you already have. You only step in when it asks a clarifying question.
No. It's being designed to be read-only: it would never write to your database, alter a table or delete a row. You'd connect it with read-only credentials you control and can revoke at any time, and you could review the SQL behind every chart.
It's designed so only schema metadata does: table names, column names, types and relationships. Row-level data would never leave your environment.
It asks. Ambiguous columns, overlapping tables and undocumented codes show up as short questions with suggested answers. Your answer becomes part of the shared definition, so nobody has to answer it twice.
A chatbot sees one question at a time and guesses. This agent keeps a persistent, reviewed model of your data, learns joins from real reports, and turns answers into charts and dashboards you can rearrange and share.
Not yet, and we'd rather be upfront about that. Plainjoin is an idea we're validating. The screens on this page are a concept preview, not a working product. Waitlist signups decide whether it gets built: if enough teams want it, we build it. Early signups help shape what it does and get first access.
Joining the waitlist is free and commits you to nothing. Pricing isn't set yet. If we build it, early signups get first access, and teams in the early pilot would use it free in return for honest feedback.
Early access
Plainjoin is an idea in validation. If enough teams want it, we'll build it. Tell us a little about yours: early signups shape what gets built and get first access.
Thanks for signing up. We'll email you as the idea takes shape, and about the early pilot. Know someone drowning in report requests? Send them this page.
Suggestion box
Plainjoin isn't built yet, so this is the best time to shape it. Tell us what you'd want it to do, or what's painful about getting reports today. Every suggestion is read by a human.
We read every suggestion, and they decide what gets built first. We'll reply to you when we can.