datatomcp
All your sheets. One place every agent can ask.
The agentic warehouse for teams without a data warehouse: your CSVs and Google Sheets, linked and kept fresh, as one remote MCP server and REST API. EU-hosted.
Upload a CSV, paste a CSV link, or pick a Google Sheet. Every import comes with a report (bad lines with their line numbers, numbers stored as text, mixed date formats), and nothing is converted until you confirm. Agents like Claude, ChatGPT and Cursor connect with one URL and a sign-in, and get read-only tools to filter, group and total your rows. Totals are computed by DuckDB, not estimated by the model, and every answer names the dataset version it came from.
Datasets that share an id or a pair of columns (country + month) are linked on import and shown on a map; agents filter, group and total across them in one query, and a linked total counts each row once. Invite your team as editors or viewers: each person connects their own agent, gets the same numbers, and an activity log shows who asked what from which agent.
Free: 3 datasets of 100,000 rows, 1 workspace, 2 people. Pro ($12 or €12 a month): 25 datasets of 2M rows, hourly sync, 3 workspaces, 5 people each.
Features
- One remote MCP server (OAuth sign-in) and a REST API over the same query engine
- Exact totals, counts and averages computed by DuckDB, never pasted into the chat
- Import report: bad lines, numbers stored as text and mixed date formats, fixed before agents see them
- Private Google Sheets through Google's file picker, with hourly sync on Pro
- Linked datasets on a relationship map, so one question can span several sheets
- Workspaces with editor and viewer roles, and an activity log of every agent question
Use Cases
- Ask Claude or ChatGPT for revenue by region on a 50,000-row sheet without pasting it
- A finance team sharing the same forecast and actuals sheets with every agent they use
- n8n, Make or your own code reading filtered totals over REST instead of downloading the whole sheet
- Revenue vs target by country across two linked sheets, in one question
Comments
Hi Fazier! I'm Killian. I help about 300 finance people use AI agents at a Fortune 100 company, and almost all of them hit the same wall: they paste a whole CSV or Google Sheet into the chat, the context fills up, and the totals are close but not exact. datatomcp is my answer to that. You connect your sheets once, and any agent (Claude, ChatGPT, Cursor, n8n) asks questions instead of reading rows. The totals are computed by a database, so the numbers are exact, and every answer says which version of the data it used. Since this week, sheets that share an id are linked, and a team can share one workspace with an activity log of who asked what. It's still early and I'm looking for honest feedback, critical included. Happy to answer any question here.
Premium Products
Sponsors
Buy
Makers
Makers
Comments
Hi Fazier! I'm Killian. I help about 300 finance people use AI agents at a Fortune 100 company, and almost all of them hit the same wall: they paste a whole CSV or Google Sheet into the chat, the context fills up, and the totals are close but not exact. datatomcp is my answer to that. You connect your sheets once, and any agent (Claude, ChatGPT, Cursor, n8n) asks questions instead of reading rows. The totals are computed by a database, so the numbers are exact, and every answer says which version of the data it used. Since this week, sheets that share an id are linked, and a team can share one workspace with an activity log of who asked what. It's still early and I'm looking for honest feedback, critical included. Happy to answer any question here.
Premium Products
New to Fazier?
Find your next favorite product or submit your own. Made by @FalakDigital.
Copyright ©2026. All Rights Reserved