Use of AI for Academic Support & Coaching

StudyTree integrates academic support services of higher education institutions with artificial intelligence to provide students with a mobile academic assistant. The software analyzes students’ grades and behavioral patterns to construct customized recommendations to improve their academic performance. Additionally, StudyTree serves advisors and administrators by providing them managerial access to the application, which enables insight to useful statistics and an overview of each student’s individual progress.

Ethan Keiser



Over the 2017-2018 academic year, two large public institutions implemented StudyTree. These universities aimed to leverage the artificial-intelligence-driven insights offered by our proprietary, original app within their undergraduate populations, and measure its effectiveness in providing learning support to that population.

The University of Washington used StudyTree in their introductory science courses to provide their students with academic coaching. LEON (the name of the AI agent) coached students by identifying problem areas, setting goals, and deploying support services. Students at the University of Washington messaged LEON throughout the academic term, ensuring constant engagement and coaching was both available and utilized.

Depending on historical conversations and LEON’s perception of the student, it was able to respond in a number of different, personalized ways. For new users, LEON responded with questions to assess the student’s perceived strengths and weaknesses, ensuring an accurate profile of the student user. Together, both the student and LEON set goals and objectives. LEON reached out to students one week before major exams to re-evaluate the student’s perception of their confidence regarding their upcoming exam. Based on this self-reflection process, LEON formed cooperative study groups, pairing students together based on strengths and weaknesses, and leveraged academic support services.

Based on work examining how educational developers collaborate across units, I recommend approaching invitations with transparency: find out about one another’s institutional roles and priorities, identify how complementary areas of expertise can be tapped for mutual benefit, and articulate how everyone will contribute. For example, educational development professionals often have well-development networks and direct connections to what is happening “on the ground” among students, instructors, and administrators, making them ideally poised to take advantage of opportunities and link efforts that might otherwise be siloed. Reciprocally, your new technology, innovation, or reimagined approach may help a center for teaching and learning reach out to a new cohort, raise useful questions about educational practices, and advance their goals for improving instruction.

Self-Reflection & Grades Outcome

Students embraced LEON as a coach and exhibited signs of an empathetic relationship. LEON helped guide students through their academic journey and connected them via formal and informal study groups. Students also opened up about their strengths and weaknesses to LEON. These self-reflections were shown to have potential predictive capability when compared

to their exam outcomes. The results from University of Washington show that LEON can be used to measure and assess student behaviors as early predictor of academic risk. LEON constantly assessed students using various coaching techniques. It proactively asked students to self-reflect on upcoming exams. 35% of students voluntarily self-reflected, providing insights into their confidence and mindset. A student’s confidence showed signs of linear relationship between the grades they received for the subsequent exams. Every 1-point increase in self-reported confidence contributes, on average 5.273 percentage increase to student grade. For final exams, every 1 point increase in self-reported confidence contributes, on average, 6.308 percentage increase to the student’s score. The analysis of student conversations with LEON and student exam outcomes suggests that there is a significant positive relationship between student self-reflection prior to the exam and the actual exam score. The study was limited in size; however, the data demonstrates student conversations with LEON are early indicator of academic outcomes. Using student behaviors to identify at-risk students, LEON can deploy effective intervention strategies and avoid poor outcomes.



Solving the resource allocation problem

The University of Maryland at Baltimore County (UMBC) offers small group and free 1-on-1 sessions to students taking 100- and 200-level courses. Students can meet with a provider, which is defined as an institution sponsored tutor, TA, coach, advisor, or any other support service staff. With over 80+ providers supporting 130 courses, the resource center staff spent most of their time handling the logistics of connecting students and providers.


Student Usage & Outcome


LEON was implemented to relieve administrative burdens and capture data on student usage of the resource center. Students scheduled appointments with providers solely through

StudyTree. A typical use case for students at UMBC was thus:

A student might seek assistance from LEON for help with a specific course through the chat interface. LEON would explore available meeting opportunities and handle the logistics of setting up a session without any administrator intervention. Over the course of one academic term, LEON successfully scheduled, completed, and recorded student feedback for 656 sessions with 48 providers and 173 students for 57 different courses. (Other courses & providers did not offer appointments). 53% of students had never used UMBC’s academic support services, indicating that LEON was capable of providing a valuable supplementary resource to the UMBC student community.

Ethan Keiser is the CEO and founder of StudyTree, an app leveraging artificial intelligence to provide students with a mobile academic assistant. He received his B.S from Drexel University in Computer Science before founding StudyTree. Ethan has won various business plan competitions including the Microsoft Imagine Cup and Dell Innovation Award, and regularly speaks about artificial intelligence at top education conferences. When not working on StudyTree, he spends time weightlifting and hiking. He is fluent in 7 languages: English and six computer languages.