Recently, we deployed Box Enterprise Advanced and have been building it into our own core technology stack. We’ve partnered with Box to explore solution development using Box Enterprise Advanced and the Box AI Studio feature set.

We’ve also partnered with mxHERO to provide intelligent email capabilities to our customers. This collaboration will lead to the creation of numerous agent candidates for industries tailored to meet the needs of users. Initially, we focused on developing a basic Intelligent Project Management Agent—a common use case that addresses the challenges project teams face.

By integrating mxHERO’s capabilities in this example, we are capturing incoming and outgoing communications in real time and applying basic workflow guardrails. The Box Intelligent Content Cloud serves as a repository for related emails and project content, allowing us to create essential metadata and bring critical project communications context and training data into the agent experience. Given the heavy communication load that project teams manage, context, team inputs and updates are key to delivering a quality agent experience.

This agent assists project managers (in this case Ralph Johnson is our PM) in organizing files, tracking project progress, and streamlining collaboration—all within the secure Box environment. It offers numerous opportunities for expansion and improvement, customized for specific tasks.

Configured using Box Content, Hubs, and Box AI Studio, we rapidly established a foundational agent experience based on a small subset of content. Initially, we are employing the default Azure OpenAI GPT-4o mini model available in the Box AI Studio, with plans for fine-tuning the agent experience post-launch and as the agent needs to evolve and deploy new capabilities.

The integration of our email communications into Box expedited the inclusion of relevant data and enhanced our data submission capabilities. Project participants can email content directly to the project manager and it captures the content in Box making it readily accessible in real time in the associated folders and in the project Hub.

As we envision program and portfolio-level capabilities for this agent to scale to support multiple projects, we have provided it with basic training data—a typical project document set stored in Box—to deliver immediate value to any project manager. The options become infinite from here.

After a rapid setup, we could now gain immediate insights based on content placed into the Project X Hub and interact with our relevant project data in a more efficient way:

1. “Good Morning, what’s on the agenda for today?”
2. “Are there any yellow or red flagged risks with Project X?”
3. “Have there been any communications about the project lately I should be aware of?”
4. “Has the customer expressed any concerns about the project?”
5. “Do I have any resources who are above their manhour time allocation plans?”
6. “Is the project on budget and on schedule?”
7. “When is the target date for the project’s launch?”
8. “Do we have a launch agenda?
9. “Are there any emails or specific files that contain CUI data we need to protect?”
10. “Create a dashboard based on the latest updates for the project with statuses.”

All of these insights are valuable with the aim of giving real-world PMs an AI assistant to help during their day, tackling the mundane.

mxHERO efficiently handles inbound and outbound emails along with their attachments, ensuring all content is well-organized and easily accessible in Box.

The result is a lightweight assistant experience when using Box AI, facilitating seamless interaction with content and related communications. The agent has content, context and can provide value to any given project manager. This experience can be further extended through APIs to integrate other relevant tools such as external data sources, LLMs and SLMs, Voice, Vision, and XR technologies.

Organizations adopting AI frequently do not have the infrastructure in place to support safe and responsible research and development, often due to internal restrictions or tight budgets. From what we are seeing, Box will provide the ability to deploy a content-based agent rapidly that provides real value from the start.

Box’s recent release has surpassed our expectations, and we are eager to continue building on this success as we continue to leverage the platform in our endeavors.