MITRE ChatGPT

The Scenario

MITRE's internal AI team had developed MITRE ChatGPT (MChat), a ChatGPT-based application, and wanted to enhance it by incorporating the company's institutional knowledge. They brought me on to help integrate MITRE data, particularly from our People Profiles system, enabling employees to find expertise in new and more powerful ways. I also designed a streamlined homepage version of MChat, which made the tool more accessible and visible, significantly increasing its adoption.

My Role

I partnered with the MChat team to strategically prioritize what MITRE data to integrate first and refine their RAG implementation to improve search results, returning more relevant employees from across the organization rather than clusters within the same department. I collaborated with the product owner and developers to evaluate different approaches for embedding MChat into our intranet homepage, weighing technical feasibility against user experience goals. I also mentored the primary MChat UX designer, providing feedback on proposals and suggestions for navigating technical constraints while building effective relationships with engineering team members. Although many of my future AI design concepts didn't advance beyond early exploration phases, they aided our stakeholders planning future AI initiatives.

Incorporating people data into the existing MChat application to find employees with relevant expertise. I wish I had saved something showing my work to help select and refine the data instead of this redacted screenshot! :-)  Source: MITRE

Exploring ideas to go beyond text-based results and integrate AI into People Profiles to better explain why someone was a match and surface expertise beyond the content on the page. Source: MITRE

My design for a streamlined version of MChat on our intranet homepage. Source: MITRE