Gatlas gives the people and agents running your companies better context.
Gatlas connects the tools each of your companies runs and describes what they hold for your agents and your team.
An agent knows only what it is handed
A group’s knowledge sits in the tools each company runs. A person knew which spreadsheet was current and which figure the accountant had restated. That knowledge was never in a tool.
More of the reading now goes to agents. Connecting an agent to one system of one company keeps getting cheaper. Reading gross margin across companies with different ledgers and currencies, by the group’s own definition, takes context that spans the group.
Context is data that is safe to act on, because its definitions and permissions travel with it. Two parts stay scarce: one definition across companies, and a boundary on who sees each one. Gatlas builds both, because an agent repeats a wrong definition at scale.
Better context has seven properties
| Property | What Gatlas does | Roadmap |
|---|---|---|
| Source | ||
| Complete | Gatlas connects the tools each company runs, and each company connects its own through a setup link. | Roadmap A company library keeps what lives in no tool: company memory, brand files and the variables everyone reads the same way. |
| Property | What Gatlas does | Roadmap |
|---|---|---|
| Curate | ||
| Relevant | You choose what each connection syncs, and your workspace decides what reaches each agent. | |
| Consistent | Gatlas defines each figure once for the whole group. You write gross margin once, and Gatlas renders it for every company and for the group. | |
| Findable | A dataset is a curated table of your companies’ data. Each one is written up with what it means and notes on its columns, for a model that has never seen your workspace. Your agent searches your files and your datasets alike. | |
| Property | What Gatlas does | Roadmap |
|---|---|---|
| Activate | ||
| Legible | Artifacts, the dashboards, documents and decks built from your data, are shaped for a person to read. Decks and documents can be kept as editions while the live artifact keeps updating. An edition keeps the data it was made from. | Roadmap Artifacts follow your own brand. |
| Delivered | The Gatlas agent answers in Slack, and an artifact posts to a Slack channel on a schedule. A reverse sync pushes a dataset into Google Sheets and keeps it current. | Roadmap Context reaches the work the moment a number moves. |
| Property | What Gatlas does | Roadmap |
|---|---|---|
| Every stage | ||
| Permitted | A member granted a company sees it and every company beneath it, and each agent connection reaches only the companies its grant names. | |
Every part of Gatlas has to improve at least one of the seven. A feature that improves none does not belong.
Six positions decide what we build
Every figure traces back to its sources.
A figure read from a dataset names the dataset it came from. A missing value shows as missing. Each dataset says when its data was last refreshed.
Defining the rows is the product.
A dataset is finished when its notes let an agent use it correctly the first time.
Each company keeps the tools it runs.
Gatlas reads each company’s tools in place, so nobody migrates.
Whoever owns a login connects it.
The owner authorizes it through a setup link, so the group never needs the credentials.
Depth comes before breadth.
Every answer inherits the definitions beneath it, so we deepen curation and search before adding sources.
A feature that works only in the dashboard is unfinished.
Your agent can build datasets, artifacts and setup links through the agentic interface, today MCP, the open standard agents use to connect to tools.
Think one of these is wrong? Send us the argument: team@gatlas.ai.
Jonathan R. FeilbergFounder and CEO, Gatlas. New York.
Gatlas started with our own companies
- Operated by
- Bifrost Partners LLC
- Built in
- New York City and Copenhagen
- First portfolio
- Our own, whose month ends gave us the first datasets and checks
If giving agents better context across a group of companies is your kind of problem, say what you would build at team@gatlas.ai.
Ask about every company from your own assistant
Create your workspace and connect your assistant. It helps you send each company a setup link for the tools it runs.