This website stores cookies on your computer. These cookies are used to collect information about how you interact with our website and allow us to remember your browser. We use this information to improve and customize your browsing experience, for analytics and metrics about our visitors both on this website and other media, and for marketing purposes. By using this website, you accept and agree to be bound by UVic’s Terms of Use for web and social media privacy.  If you do not agree to the above, you can configure your browser’s setting to “do not track.”

Skip to main content
People Departments Maps Buildings Rooms A-Z
University
of Victoria
Apply Library
Sign in Online tools Sign out
Faculty of Graduate Studies

Nitin Gupta

  • BEng (University of Victoria, 2022)

Notice of the Final Oral Examination for the Degree of Master of Science

Topic

Context-Aware AI for Nonprofit Email Communication: An MCP-Based Approach

Department of Computer Science

Date & location

  • Monday, April 27, 2026

  • 5:30 P.M.

  • Virtual Defence

Reviewers

Supervisory Committee

  • Dr. Daniela Damian, Department of Computer Science, University of Victoria (Supervisor)

  • Dr. Neil Ernst, Department of Computer Science, UVic (Member) 

External Examiner

  • Dr. Christoph Treude, Department of Computer Science, Singapore Management University 

Chair of Oral Examination

  • Dr. Daniela Constantinescu, Department of Mechanical Engineering, UVic

     

Abstract

The Model Context Protocol (MCP) enables large language models to dynamically integrate external organizational context, supporting more context-aware AI applications. We investigate how context-aware AI approaches can support nonprofit email communication workflows, and in particular how the amount of contextual information available at generation time influences staff perceptions of the quality and usability of AI-generated responses. To address this question, we introduce the Knowledge Aware Nonprofit Tooling Architecture (KANTA), an MCP-based design that connects AI models to nonprofit data and tools, including mission statements, conversation history, and user-uploaded context. In an empirical evaluation with six nonprofits and eight representatives (22 scenarios), participants evaluated three context variants per thread: light (mission only), medium (mission plus current thread), and full (adds email history and uploaded context). They rated clarity, tone/personalization, and likelihood-to-use. Across measures, full-context replies received higher satisfaction ratings than light and medium. A linear mixed-effects analysis indicated significant improvements for full over lighter variants, and a nonparametric test showed a statistically significant effect favoring full-context responses. Qualitative analysis showed participants perceived full-context outputs as “knowing what it’s talking about,” requiring less editing, and better matching relationship tone; suggestions emphasized asking clarifying questions when information was missing and maintaining professional formatting. We discuss how tool-mediated contextual grounding can be integrated into existing nonprofit workflows, why combining it with interactive clarification and, where appropriate, retrieval augmentation may further improve usability, and limitations of this exploratory sample. Our results offer early evidence that richer organizational context can measurably improve the perceived quality and usability of AI-generated nonprofit emails.

Back to oral exams