We put a whole portfolio’s data behind one MCP server
Gatlas is a portfolio data platform built in New York City and Copenhagen, operated by Bifrost Partners LLC. It syncs every company’s tools into documented, context-efficient views and serves them to AI agents over MCP.
17 live sources syncing portfolio tools.
0 credentials passing through you.
Our Mission
An agent is only as good as the data it can reach, and across a portfolio, that data lives in a dozen tools, in a dozen formats, behind a dozen logins.
Our mission is to make all of a portfolio’s data available through one MCP server, organized to be context-efficient, so an agent reads the few documented rows that answer instead of raw tables.
Core Values
The principles that guide every decision we make, from schema to shipping.
Truth Over Plausibility
A wrong number that looks right is worse than an error. We would rather fail loudly than serve a figure we can't trace back to the books.
Craft & Care
Details matter. A view isn't done when the SQL runs; it's done when its AI context and column docs let an agent use it correctly.
Build With Users
Gatlas is shaped by the questions real operators ask. Their day-to-day usage drives the error hints, views, and reference data we ship.
Governed by Default
Access is scoped to entities, views, and tools; read-only by default; every query logged. Trust in the answers starts with control over who can ask what.
- Every Query Logged
- Each MCP call is recorded, so agent access to the books stays auditable.
- One MCP Server per Org
- The whole portfolio answers from a single endpoint, for Claude or any MCP client.
- Contact
- Write to team@gatlas.ai and a person reads it.
- 0 Credentials Collected
- Operators connect their own tools through delegated setup links.
- Built in New York City and Copenhagen
- Gatlas is operated by Bifrost Partners LLC and built in New York City and Copenhagen.
- 17 Live Sources
- From accounting to analytics, synced into one semantic layer.
The Team
Gatlas is built by a deliberately small team in New York City and Copenhagen, operated by Bifrost Partners LLC, and we run our own portfolio on it every day.
The product exists because we needed it. Every connector, view, and guardrail was shaped by real month-ends, real reconciliations, and real questions asked of real books.
Work with us
We don’t keep a list of open roles here. If making a portfolio’s data agent-ready sounds like your kind of problem, we’d like to hear from you.
Tell us what you’d want to build: write to team@gatlas.ai.