Box AI API is available to all customers Business and above.

Get a summary

Summarize a document in a selected tone of voice.

Demo

Extract metadata from files (freeform)

Use natural langage, or pass a stringified data structure to extract metadata with Box AI API.

Demo

Extract metadata from files (structured)

Extract document metadata suggestions. Pass a Box metadata template ID or a predefined data structure.

Demo

Use Box AI Enhanced Extract Agent

Extract document metadata suggestions with the new Enhanced Extract Agent. Get the best answers thanks to chain-of-thought processing.

New

Tutorial

These resources will get you up and running with Box AI API.

Box AI API reference

Check the API reference for specification details.

Documentation

Get started with Box AI API

Browse the guides to learn how to use Box AI API.

Documentation

Supported AI models

Box supports a variety of AI models, categorized along two dimensions: access level and capability tier. Check table list of the supported AI models.

Documentation

Override AI model configuration

Override the default AI model configuration and fine tune LLMs with Box AI API.

Documentation

These resources will get you up and running with Box MCP server.

Remote Box MCP server

The remote Box MCP server allows third party AI systems to securely connect and interact with your content in Box.

New

MCP

Self-hosted Box MCP server

A Python Developer Community open source project. It integrates with the Box API to perform various operations such as file search, text extraction, AI-based querying, and data extraction.

MCP

Build AI Apps with MCP Servers

Join a short online course, built in partnership with Deeplearning.ai, and taught by Ben Kus, Box CTO.

Short Course

Box MCP server and Pydantic AI

Use Box Doc Gen through the Box MCP server for AI-Powered Document Generation.

Tutorial

Box MCP server and LangChain MCP Adapters

Turn Box MCP server into a LangChain-compatible agent using the LangChain MCP Adapters.

Tutorial

Learn how to create advanced AI agents with Box.

Manage agents with Box AI Studio API

Create and manage custom AI agents with Box AI Studio API.

New

Box AI Studio

Box for Agentforce Extension package

Get started with reusable Agentforce actions that help automate workflows and enhance intelligent agent—based processes within Salesforce.

New

Salesforce

Multi-agent workflows with Box and OpenAI

Use OpenAI Responses API and a Box Agent to add unstructured data from your Box instance into your agentic workflows.

OpenAI

Box AI Agents Toolkit

Get started with a Python library for building AI agents for Box.

Beta

Use Box for AI Integrations to extend LLM models' existing knowledge bases.

LangChain.js

Include Box content in your LLM workflows with Box loader for LangChain.js.

New

Integration

Openflow Connector for Box

Seamlessly connect unstructured content in Box with the powerful analytics capabilities of Snowflake, unlocking new insights and automating data-driven workflows.

Integration

Airbyte

Transform unstructured documents into structured, queryable data by using the “Box data extract”, an Airbyte source connector.

Integration

Weaviate

Build RAG workflows by embedding Box content into a Weaviate vector database and leveraging Weaviate's new Query Agent.

Integration

Pinecone

Connect Box and Pinecone to customize vector embeddings and get more relevant answers from LLM.

Integration

LlamaIndex

Enable access to Box content within LLM workflows with Box reader suite for LlamaIndex.

Integration

LangChain

Include Box content in your LLM workflows with Box loader for LangChain.

Integration

Pydantic AI and Box MCP

Use the Box MCP server to augment Pydantic AI agents with secure content in Box.

Demo

Watch the latest Box AI API tutorials and demos.