Ask. Don't buy.

Software at the speed of a question. Describe the app your team needs. It is live in minutes, on real data, always up to date, shared with your team.

The person who builds and connects pays for one seat, EUR 25/month. The people who only read the app do not pay for a seat. The public data catalog is included.

Apps that already run here

Each one is a data app: it answers a question with real numbers, on a URL a team opens.

Built by autario, on autario, through the same public API and the same MCP tools an outside builder uses.

Bubble or Not: Stock price against revenue, earnings, and shares for any company. Spot a bubble from the gap between price and fundamentals.
Screener: Rank and filter companies across fundamentals: revenue growth, margins, valuation, all from primary filings.
Chart Builder: Turn any dataset into a publishable chart in seconds. The first app built on the autario data layer, dogfooding the public API.

Your app is next. Describe it to your AI, publish it, and your team opens a link.

Browse all apps

How an app gets here

No canvas, no widgets to drag, no procurement round.

Describe the app your team needs, in your own words.
Connect the sources it runs on. We hold the OAuth apps, the refresh and the schema changes.
Your AI builds it against normalized data, through MCP or the SDK.
It goes live on a URL, always up to date, on a schedule or on demand.
Share it with your team. The people who only read it do not pay for a seat.

What one team pays

One person builds and connects. Everyone else opens a link.

What you pay for	autario
The person who builds and connects	EUR 25/month
The people who only read the app	No seat, no charge
Connectors, OAuth apps and refresh	Included
Hosting, the URL and the scheduler	Included
The public data catalog	Included

Per seat pricing was the right model while building software was hard. Building is no longer the hard part.

Who builds it

Same data layer, same public API, three kinds of builder.

Your AI builds it: Connect autario as an MCP server and let Claude, ChatGPT or your own agent read the data and ship the app. Connect via MCP
You build it: A one file SDK, a gated data bridge and hosting included. Write the app yourself and we run it. Read the developer docs
You click it together: Pick a dataset, pick a shape, publish the chart. No code at all. See the apps

Ask for the answer, not the table

One filtered call returns the rows you asked for. Pasting the source table into a prompt returns everything else with them. Both sides below are measured, not estimated.

Ask

What was the unemployment rate in Germany from 2010 to 2020?

query_dataset(dataset_id, filter: ["countryiso3code:eq:DEU", "date:gte:2010", "date:lte:2020"], sort: "date:asc")

The same call over REST:

GET /api/v1/public/datasets/6217e2ba-f66a-45c9-8d0d-953daeb0670d/data?filter=countryiso3code:eq:DEU&filter=date:gte:2010&filter=date:lte:2020&sort=date:asc

Answer

Year	Value (%)
2010	7.043
2011	5.967
2012	5.372

First 3 of 11 rows.

11 rows, about 356 tokens in TOON.

Pasting the same period for all 265 entities instead: 2,915 rows, about 181,731 tokens in JSON.

Roughly 510 times fewer tokens for the same question. Comparing the same wire format on both sides, the gap is about 254 times.

Measured on 2026-09-04 against Unemployment Rate by Country (World Bank), published by World Bank. Token counts are approximations at about four characters per token.

Source dataset: Unemployment Rate by Country (World Bank)

The public data catalog comes with the seat

A marketplace of verified datasets, every one linked to its primary source with no hallucinated numbers. Cleaned, schema-detected, and entity-resolved into a shared ontology at ingestion. Filter by topic, source, or region. Download as CSV or query live via API.

Trusted sources. Real data. No hallucinations.

Normalized at ingestion: Every dataset is cleaned, schema-detected, entity-resolved and shadow-columned so cross-dataset joins feel like magic.
AI-native: MCP server for any AI agent. LLMs search, query, and publish charts with verified data. No training-data contamination.
Always fresh: A self-healing ingest engine refreshes sources nightly. Full-replace strategy means you always see the latest numbers.
Open by default: Public datasets are free. No login to browse or query. API keys only for publishing. Like GitHub, but for data.
Every claim verifiable: Sources from World Bank, IMF, Eurostat, OECD, WHO, FRED, US Census, SEC and more. Each dataset links to its primary source.
LLM-agnostic: No provider lock-in, no "Powered by X". Connect Claude, ChatGPT, Gemini, or any agent that speaks MCP.

Start with a question

Open an app, read the data, or connect your AI through MCP. Looking costs nothing and needs no account.