Skip to content

Portfolio intelligence, in a form your agents can use.


Gatlas turns what each company’s systems hold into documented datasets and serves them to your agents over MCP. Ask across the whole portfolio in plain language, in Claude or in Slack, and every answer traces back to its source.

Gatlas reads from the tools your portfolio already runs on: accounting, banking, commerce, CRM, HR, analytics and the files in shared drives

Shopify logoShopify
Stripe logoStripe
QuickBooks logoQuickBooks
e-conomic logoe-conomic
Xero logoXero
Google Sheets logoGoogle Sheets
Google Drive logoGoogle Drive
GA4 logoGA4
Supabase logoSupabase
Attio logoAttio
HubSpot logoHubSpot
Mercury logoMercury
SharePoint logoSharePoint
Microsoft Teams logoMicrosoft Teams
Power BI logoPower BI

Three steps to one MCP server

Connect every company’s tools, model the data into documented datasets, and let your agents ask, governed from the first query.

Connect

Delegated setup links: operators connect their own tools, and no credential passes through you.

Model

Each dataset carries a description written for a model and notes on its columns, so your agent reads the few rows that answer, not raw tables.

Ask

Ask from Claude or any MCP client, or add the Gatlas agent to Slack. Each connection reaches only the companies you choose, and every query is logged.

Answers where your team works

Add the Gatlas agent to Slack and your team asks there. It answers from your Gatlas data, read-only, and names the dataset behind every figure. Artifacts post to a channel on the schedule you set.

Your best queries become artifacts

Query raw and modeled data in the built-in SQL workspace, and save the queries worth keeping as datasets. Artifacts lay them out in the form a reader knows, from a dashboard to a deck, built from components you change once and see change everywhere.

Every company rolls up into one answer

Roll results up across companies and currencies: one query spans the whole portfolio.

Each company stays on its own

Ask about one company, see only that company. Each company’s data lives in its own schema, and each MCP connection reaches only the companies you choose.

Syncs on a schedule

Every connection syncs on its own schedule, and you can trigger a fresh run any time, from the dashboard or over MCP.

Files sit next to the numbers

Sync board packs and contracts from Google Drive, SharePoint and OneDrive. Your agent searches the files and reads any one in full, and a prompt distills them into short documents beside your datasets.

Run the whole portfolio from one place

Five supporting pieces that carry data from each company's tools to answers you can act on.

Ask in Claude, answer from the books.

Connect Claude or any MCP client to Gatlas, or add the Gatlas agent to Slack, and ask across every company in plain language.

Get started

Connect the tools your companies already use

Sources across accounting, banking, commerce, CRM, HR, analytics and shared drives: Shopify, Stripe, QuickBooks, e-conomic, Xero, HubSpot, SharePoint and more. Send an operator a setup link and they pick and connect their own systems, so no credential ever passes through you.

Shopify
QuickBooks
Xero
Attio
Stripe

Governed by default

Give agents real access to the numbers without giving up control of them.

Scoped connections. Every MCP connection reaches only the companies you choose. An agent sees those companies and no others.

Scoped connections, each Read-only or Read/write, and a log of every query.

Your agents can ask about every company

Sync every company's tools into documented datasets and give your agents one MCP server to ask.