Skip to main content
The Overten API lets you generate professional Office documents by describing what you want in natural language. The same agents that power the Overten web app produce .xlsx, .docx, and .pptx files your users can download, embed in an iframe, or keep iterating on — all from a single API call.

Generation endpoints

Each endpoint accepts an output_formats array, so you get every shape you need in one request. A single Word call can return both an editable .docx and a print-ready .pdf:
EndpointSupported output_formats
POST /excel/generatexlsx
POST /word/generatedocx, pdf
POST /slides/generatepptx, pdf
The response includes one download_url per requested format and a run_id you can pass back to resume the same document on the next turn.

What you can build

Report generators

Feed numbers and a prompt, receive a polished spreadsheet with charts, formulas, and conditional formatting.

Deck builders

Turn a sales brief into a 10-slide investor deck with cohesive branding.

Proposal automation

Merge template prose with dynamic client data into well-formatted Word documents.

Document analysis

Ask natural-language questions of an existing Excel file and get structured answers back — without producing a new file.

How it works

Every generation follows the same five-step lifecycle:
1

Sign up

Exchange a Firebase ID token for an sk_live_* API key via POST /signup. See Authentication for the full flow.
2

Upload reference files (optional)

Upload any reference files or images you want the agent to use as context before generating.
3

Generate

Call POST /excel/generate, /word/generate, or /slides/generate with a prompt. The agent plans, builds, and returns a finished file.
POST /excel/generate
{ "prompt": "Create a Q3 sales report with summary statistics" }
4

Poll or receive a webhook

In sync mode the finished document is in the HTTP response. In async mode you receive a task_id immediately and poll GET /tasks/{task_id} or register a webhook_url to be notified when the run finishes.
5

Download or embed

Each completed run returns download_url values for every requested format. Pass the edit_url to the Overten embedded editor if you want users to keep editing in your app.
Every generation endpoint returns a run_id. Pass it back on the next call to resume — the agent keeps its memory of what it built, so “now add a forecast tab” costs a fraction of what turn 1 did.

Sync vs async

Every generation endpoint accepts async: true. Sync mode waits up to 300 seconds and returns the finished document in the HTTP response — ideal for short jobs and prototyping. Async mode returns a task_id immediately so your server stays free while the agent works. See Async vs sync for guidance on which to use.

Status and support

Where to go next

Quickstart

Make your first API call and download a real document in under five minutes.

Authentication

Sign up and mint an sk_live_* key in one request.

API reference

Every endpoint, request schema, and response shape.

Changelog

Release notes and deprecations.