Chart a private CSV with an AI agent.

What is the direct answer to this workbook question?

To chart a confidential CSV without uploading it, open the file in xmetric in your browser and share only the column headers with your AI agent. The agent writes the formula; you build the chart on shelves in the xmetric app. CSV, Excel and Parquet files are read locally and nothing is uploaded unless you save online.

Why do cloud data uploads complicate privacy reviews?

When you drop a customer export or payroll spreadsheet into ChatGPT or Julius AI, the entire file travels over the network to a remote cloud workspace. For finance, healthcare, and customer PII exports, uploading raw rows violates internal security policies.

Chat-based Python sandboxes also usually return a static image or need another prompt every time you want to filter a date range, swap a dimension, or drill into an outlier.

How can I share schema and keep the file in my browser?

You can get the speed of an AI assistant without sending your rows off your computer:

1. Open your CSV, Excel (.xlsx) or Parquet file in the xmetric app (/app/) in your browser. The file stays in local browser memory.
2. Send only column names and data types to your AI agent when asking for a formula.
3. Paste the formula as a calculated field, test it in the formula playground (/play/) if you like, then place fields on Rows, Columns, the Marks card and Filters to build the chart.

head -n 1 customer_invoices.csv

The public xmetric MCP server is live at https://getxmetric.com/mcp with no authentication or API key. Anonymous initialize, tool discovery and widget-resource checks passed on 2026-10-29. ChatGPT and Claude host rendering and tool selection remain unverified; their directory listings remain coming. The shipped tools expose bounded schema and aggregate queries plus specific worksheet commands; they do not promise arbitrary formula authoring.

How can I check cohort retention from a private CSV?

Suppose your local invoices.csv has three columns: customer_id, invoice_date, and amount_paid. You want to chart monthly revenue by the customer's signup cohort without uploading the customer list.

Ask your agent for the Level of Detail formula using only those three column names, then test it in the xmetric formula playground:

// 1. Cohort Month (Calculated Field in xmetric)
DATE_TRUNC('month', { FIXED [customer_id] : MIN([invoice_date]) })

// 2. Months Since First Invoice
DATEDIFF('month', [Cohort Month], DATE_TRUNC('month', [invoice_date]))

-- Equivalent SQL
WITH cohorts AS (
  SELECT
    customer_id,
    invoice_date,
    amount_paid,
    DATE_TRUNC('month', MIN(invoice_date) OVER (PARTITION BY customer_id)) AS cohort_month
  FROM invoices
)
SELECT
  cohort_month,
  DATE_DIFF('month', cohort_month, DATE_TRUNC('month', invoice_date)) AS months_since_join,
  SUM(amount_paid) AS cohort_revenue
FROM cohorts
GROUP BY 1, 2
ORDER BY 1, 2;

In the xmetric app, place Cohort Month on Rows, Months Since First Invoice on Columns, and SUM(amount_paid) on Color to get an interactive cohort heat map in your browser.

How do cloud analysis and local browser BI differ?

A cloud analysis flow sends data to the host; a browser-local flow can keep the original file local while you choose which metadata or results to share.

When are ChatGPT or Julius AI a better fit?

Choose ChatGPT Data Analysis or Julius AI if your dataset is non-sensitive and you need arbitrary Python libraries like scikit-learn for machine learning regression, PDF text scraping, or automated multi-step Python notebooks in the cloud.

When should I use xmetric?

Choose xmetric when your data cannot leave your laptop and you want to build charts on shelves, write Level of Detail expressions and table calculations, or inspect workbooks for free in your browser.

What is live, and what remains planned?

The public xmetric MCP server is live at https://getxmetric.com/mcp with no authentication or API key. Anonymous initialize, tool discovery and widget-resource checks passed on 2026-10-29. ChatGPT and Claude host rendering and tool selection remain unverified; their directory listings remain coming. See the dated public protocol receipt and setup and evidence limits.

The public server is available; host operation and directory approval are separate checks.

The file board parses supported .pbix, .twbx/.twb, CSV, Excel and Parquet files in your browser. Choosing a file shares supported schema, source names and calculation definitions with xmetric and the chat host; requested result cells, filters and SQL or DAX evidence are shared when answering. The original file is not uploaded by this flow. Small aggregates and labels can still be sensitive. Privacy details.

What data does this worked example use?

The initial view uses authored synthetic data, not customer data or a captured agent session.

The example record lists its file, calculation or SQL and expected values. The live browser result is computed by xmetric’s existing engine or workbook inspector. Samples are excluded from own-data activation reporting.