DataTugPull more from your data
Open app
☰

For analysts, new team members and AI agents

Understand what your data means

Column names do not say what data means. DataTug records meanings once for the whole project; MeaningGraph, planned, is to link them across sources so people and agents read the same definitions.

Explore the demo project All five solutions

Starts here: AI agent buildersData analysts

Also useful: Developers (also useful)Helpdesk (also useful)SRE and on-call (also useful)

Watercolor of a cartographer’s hands joining an old sepia map and a modern blue chart of one coast with a thread between the same lighthouse on each.

Illustration · sample data, not a capture

One order, three column names

The same order in three systems: the column that holds it, what it means, and how the meaning is recorded
WhereColumnMeaning
NorthwindOrders.OrderIDOrder ID
Paymentspayments.order_refOrder ID
Carrier APIrequests.order_idOrder ID
  • A person records each meaning once, for the project. Column meaningsAvailable
  • MeaningGraph links the three as the same order, so a join needs no ON clause. MeaningGraph integrationPlanned

Static illustration of the intended flow, not a capture of the product. Column meanings are available; MeaningGraph linking and join-by-meaning are planned.

Watercolor of two dockers holding identical tags but pointing at different ships, while the old harbour pilot who knows dozes in a chair.

The problem

The order is OrderID in one database, order_ref in another and order_id in the carrier’s API. An amount is cents in one table and dollars in the next. A status of 3 means “packed” to the people who wrote the code and nothing to anyone else.

That knowledge lives in a few heads. Every analyst writes a private glossary, and an AI model asked to join two tables guesses from the names.

Before and after

Who: Analysts, new team members and AI agents

Today

4 tools2 hand-offsno shared password

  1. Analyst, waiting for a colleague · ChatAsk the one person who knows, and wait.
  2. Developer · Code editorRead application code to learn what a status value means.
  3. Two colleagues · SpreadsheetKeep a private glossary that disagrees with a colleague’s.
  4. AI agent, then a person · SQL clientLet an agent guess a join, then check it by hand.
Today4 tools · 2 hand-offs · no shared passwordAsk the one person who knows, and wait.Read application code to learn what a status value means.Keep a private glossary that disagrees with a colleague’s.Click to see

With DataTug

1 workspace0 hand-offsno shared password

Some steps are planned or in development.

  1. Anyone on the projectBrowse the schema of every source in the project in one place.Schema browsing in the web appPreview
  2. Whoever knows, onceGive a column a meaning (Order ID, Email, Amount in cents) once, for the whole project.Column meaningsAvailable
  3. MeaningGraphMeaningGraph links the same meaning across sources, so the order is recognised in every database that holds it.MeaningGraph integrationPlanned
  4. A person confirmsMeanings are suggested from how the team queries, and confirmed by a person.MeaningGraph learning loopExploring
With DataTug1 workspace · 0 hand-offs · no shared passwordBrowse the schema of every source in the project in one place.Give a column a meaning (Order ID, Email, Amount in cents) once, for the whole project.MeaningGraph links the same meaning across sources, so the order is recognised in every database that holds it.Click to see

Each step names the part of DataTug it relies on and that part’s status today. “Planned” and “In development” steps describe where the product is going, not what you can use now.

Where it meets the other solutions

Try it on sample data

The public demo project opens in the browser with sample Northwind and Chinook data. No sign-up.

Explore the demo project