> ## Documentation Index
> Fetch the complete documentation index at: https://docs.numi.sh/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Revenue by business segment, product and region for US SEC filers, each breakdown checked against reported revenue, and fundamentals as filed.

The numi API serves company data from filings with the US Securities and Exchange Commission (SEC):

* **Segments by axis:** revenue by business segment, by product and by region. Each breakdown is an axis, and each axis is its own list with a `reconciliation` status: `reconciled` when its rows add up to the revenue that the filing reports.
* **Fundamentals as filed:** revenue, income, balance sheet lines, cash flow and earnings per share, each with the XBRL concept, unit and period that the filing reports.
* **Segment notes:** the note in which the company describes its segments, as plain text.
* **Bulk files:** every table of the API (companies, filings, axes, segments and line items) as Parquet and as CSV.

There are three ways to get the data:

| Interface | Address | For |
| - | - | - |
| REST API | `https://api.numi.sh/v1` | Programs and scripts |
| MCP server | `https://api.numi.sh/mcp` | Agents such as Claude, ChatGPT and Cursor |
| Bulk files | `GET /v1/bulk/latest` | Loading every table into a database or a notebook |

## Companies and forms

The data is from US SEC filers whose filings numi has collected, and from these forms: annual reports on Form 10-K, 20-F and 40-F, quarterly reports on Form 10-Q, and the amendments of each. The number of companies in each release is in the [changelog](https://docs.numi.sh/changelog.md).

On the Builder plan, the paid plan, you can ask numi to extract a company that is not in the data yet. See [on-demand extraction](https://docs.numi.sh/plans.md#on-demand-extraction).

## Filed facts and computed fields

Each figure is a fact that a filing reports. The API computes two things from those facts: the share of revenue of each row (`revenue_share`), and the reconciliation of each axis (`segment_revenue_sum`, `reconciling_revenue_sum`, `coverage_ratio` and `reconciliation`). Nothing is estimated, restated or converted from one currency to another.

> **[Quickstart](https://docs.numi.sh/quickstart.md)**
>
> An API key and the first call

> **[Segments](https://docs.numi.sh/segments.md)**
>
> Axes, rows and reconciliation

> **[MCP server](https://docs.numi.sh/mcp-server.md)**
>
> Connect an agent

> **[Bulk files](https://docs.numi.sh/bulk.md)**
>
> Every table as Parquet and CSV


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.