Potential knowledge gaps for first-generation students may include such topics as:
- Financial aid and billing. Many students need additional guidance regarding aid eligibility, renewal requirements and completing the Free Application for Federal Student Aid. Limited familiarity with billing processes may result in additional fees or temporary loss of access to university services.
- Course registration policies. Rules surrounding course adds and withdrawals are often misunderstood and can carry unintended academic or financial consequences.
- University office functions. Students often struggle to distinguish between the types of services that are provided by different central offices.
- Navigating the physical campus environment. Uncertainty about office locations, building names and parking regulations can create barriers for students seeking in-person support.
- Academic terminology. Higher education jargon is not always self-explanatory, which creates access barriers for those with limited prior exposure.
Creating AI Tools for Equity-Focused Advising
Common success barriers can inform targeted AI-driven advising tools that offer first-generation students on-demand guidance as challenges arise in real time. Most major AI platforms allow users to create specialized agents that can be configured to perform specific tasks using a defined knowledge base. Remember that these tools are intended to supplement, not replace, human advising relationships. A background in computer programming is not necessary to personalize AI agents. Although platforms differ slightly, the process involves only a few basic steps.
WATCH: Four AI trends to keep an eye on.
Provide clear instructions. Think of these instructions as if you are describing job responsibilities to a new employee. Because these platforms are language-based models, you can describe the agent’s role explicitly in sentence form. In the instructions, specify the agent’s purpose, who it serves, and how it should communicate with the end user. Provide clear task boundaries, describing topics that it should or should not respond to, and when it is appropriate to trigger a referral to a human student adviser.
Upload a knowledge base. Provide the agent with documents that it will draw knowledge from. Include documents such as frequently asked questions or policy guides that the agent will use in its responses to students. Remember to include some of those “unspoken” campus norms that first-generation students may be unfamiliar with, including common colloquialisms relevant to student life.
To optimize the agent’s functionality, keep documents concise and well organized so the system processes less information to generate responses. While uploading dense handbooks may not be entirely useful, creating PDFs with pointed information organized by topic is a more viable starting point. Note that most platforms also allow you to choose whether to enable internet searching as part of the agent’s capabilities, which provides greater control over knowledge sources.
Test and refine the agent. Have several staff members test the agent for accuracy and functionality by asking anticipated student questions. This is a great opportunity to enlist the help of student employees who may be better equipped to generate questions and provide feedback on the agent’s performance. Use their feedback to refine the agent’s instructions and knowledge sources until it appears to be functioning optimally. Note any AI hallucinations and mitigate them through explicit constraints in the agent’s instructions.
DISCOVER: Google Gemini can improve learning in higher ed.
Decide who owns it. Depending on your institution’s unique support of AI platforms, the custom agent may need to be housed under a user’s profile rather than an entity account. This point is worth considering given that the agent’s instructions and knowledge sources will require updates as policies and procedures evolve. Practically speaking, institutions should designate an individual to manage the agent and the accompanying documents.
Share it. The appearance of your AI agent should instill student confidence, so be sure to give it a clear name and description that conveys its purpose. Before sharing with students, be mindful of the access settings you select. Most platforms allow you to choose a link that is private, public or available to anyone with organizational access, which can be embedded in university websites and communications. Ultimately, this decision should be made based on data sensitivity levels and ease of student access. Even if students are not entering sensitive information, institutions have strict policies governing the sharing of intellectual property that must be considered.
With a growing demand from students and university administrators for AI integration in higher education, custom agents offer an accessible entry point for educators and staff to bring AI innovation into their work. Smaller, specialized student populations, who often benefit most from supplemental advising resources, present a natural opportunity to pilot these tools before scaling them more broadly.
Now is the time for higher education professionals to leverage their creativity and commitment to student equity by engaging with emerging technologies that have the potential to reshape how we support students moving forward.
