Building a multi-agent workflow with OpenAI’s Agent Builder + Box MCP server

See how to build a multi-agent workflow using OpenAI's Agent Builder and the Box MCP server. This demo shows you how to construct an automated client onboarding process that verifies documents, extracts key data, and drafts personalized notes securely within Box.

If you are looking to build agentic applications, this video provides a direct look at how to connect OpenAI's agent framework to Box. We show you how to orchestrate tasks where all content processing happens inside Box, ensuring data remains secure and governed.

Key takeaways:

→ Build a three-agent workflow in OpenAI's new Agent Builder
→ Connect agents to the Box MCP server to access secure content tools
→ Use conversational prompts to trigger complex, multi-step processes
→ Add conditional logic between workflow steps
→ Automatically generate and save new files back into a Box folder

Chapters:00:00 - Introduction

00:55 - Designing the agentic workflow
01:51 - Connecting Agent Builder to the Box MCP server
03:06 - Configuring workflow logic and state
03:42 - Running the automated onboarding flow
05:06 - Reviewing the AI-generated output in Box
05:50 - Getting started with the Box MCP server

Follow us on social media:
X: https://www.x.com/box
LinkedIn:   / box  

Get started with the Box MCP server:https://blog.box.com/box-enables-agen...