Getting started

aqmen is the platform a strategic decision is built on. A workspace holds the data with its sources, the transformations that derive from it, the spreadsheet models, the charts, the views and the conclusions — connected, so every number traces back to a document and anything that goes out of date is flagged. You can do all of it in the app; an agent connected over MCP does most of it faster, and leaves the same trail.

1. Sign in and create a workspace

Sign in and create a workspace. Check Start with demo data to get a complete worked example: a bottom-up HR-software market for seven EU countries, built from Eurostat tables and vendor prices with every row cited. It is the fastest way to see how the pieces fit, and you can delete it any time.

2. The pieces

The sidebar lists the workspace by kind:

• Datasets — tables, imported from CSV, JSON or Parquet files or loaded by an agent. Every dataset carries sources (citations) and every column its meaning: type, unit, scale, description.
• Transformations — SQL that derives a new dataset from others. The SQL is the method; the output is a dataset like any other.
• Spreadsheets — Excel-compatible workbooks whose tables are connected to datasets, with calculated columns, checks and versions. Where models live.
• Charts — a SQL query or a spreadsheet range with a picture. They re-run when opened, so they always show the current data.
• Views — small pages composed from charts and ranges, one per question, written by an agent from your description.
• Insights — one conclusion pinned to the chart or spreadsheet that shows it, flagged when the data behind it changes.
• Lineage — the graph from sources to insights.
• Activity — who did what, in order, agents included.

Collections are folders across all of these. The workspace Overview shows everything as a file tree (Explore) with the collections as folders, and a Docs tab for the workspace README.

3. Load data

Datasets → Import data: drop a file, check how it parses, describe the columns, add the sources, import. See Import & sources.

4. Derive, model, chart

Write a transformation when a dataset needs deriving (Transformations → New). Build a chart from a query or a spreadsheet range (Charts → New). Ask an agent to build the spreadsheet model and the views. See the pages that follow.

5. Find anything

⌘K opens the command palette: every entity, page and action in one box. Each entity page has an About rail with its lineage — what it reads and what reads it — and a Docs tab for notes that agents read.

6. Connect an agent

The recommended way to work. Settings → MCP integration shows the URL to paste into Claude, ChatGPT or any MCP client. See Connect an agent.

Faster with an agent — once connected, your first prompt can be as simple as "describe my workspace and tell me what's worth looking at." The agent reads the datasets, their meaning, the models and the recorded insights, and comes back oriented. Every page of this guide ends with a prompt like this one.