AI Enrollment

Automating Transcript Evaluation for Higher-Ed Admissions

At EdVisorly, we set out to tackle a complex challenge that burdens nearly every admissions and enrollment office in higher education: processing student transcripts. Every transcript arrives in a different format — different layouts, course codes, GPAs, rigor classifications, weighting systems, and sometimes even handwritten notes. For admissions teams, especially during peak season, reviewing these documents takes time away from what matters most: engaging with students.

The status quo included manually extracting course and grade information into spreadsheets, recalculating GPAs by hand, cross-checking requirements, and uploading cleaned data into CRMs such as Slate or Salesforce. Institutions like Stony Brook University saw 73% growth in undergraduate applications over five years, yet they were trying to manage that volume using the same staffing levels. Texas Tech University was spending 2–5 minutes per transcript, doing manual GPA computations and searching for duplicates. Multiply that by tens of thousands of applicants, and you have teams burning countless hours on clerical tasks — instead of reviewing students holistically.

EddyAI™ was created to automate this workflow.

The Challenge

When I joined this initiative, there were two parallel needs:

1. Enrollment teams needed automation that drastically reduced manual processing time.

2. Adoption could only succeed if the user experience built trust in the AI.

This wasn’t just about machine learning; it was about designing a system that helped humans make decisions faster and with more confidence — not replacing staff or hiding the process behind a black box.

My responsibility was to lead the design from the ground up: research, product definition, UX strategy, and visual UI design, while collaborating with AI engineers and university enrollment teams to ensure we were solving real problems, not theoretical ones.

Research & Discovery

I began by interviewing admissions and enrollment operations teams across multiple universities. We walked through their current process step by step:

• How do transcripts arrive?

• What data must be extracted?

• Where do manual errors occur?

• When do files get stuck or misrouted?

• What decisions require human judgment?

I mapped these observations into process flows and user journey maps — capturing real bottlenecks like inaccurate OCR tools, inconsistent course coding, duplicate transcripts, and the dreaded “spreadsheet handoff.”

One key insight shifted the direction of our product:

The goal wasn’t just automation — it was transparency and control.Admissions teams needed to see what the AI extracted, validate it quickly, and override it if needed.

That insight shaped the core design philosophy: AI-powered, human-verified.

Design Approach

From there, I moved into prototyping. The UI needed to support high-volume, repeatable work while making transcript validation feel light and intuitive.

I designed a dashboard that shows:

• transcripts flowing in,

• the extraction status of each file,

• confidence levels on data accuracy,

• flags where human review is required.

Every detail was designed to reduce cognitive load — from data grouping to typography hierarchy. Teams could scan, validate, and approve transcripts with minimal clicks.

AI engineers and I worked closely to translate model outputs into interface patterns. For example, we introduced confidence scoring to visually show how certain the model was about a course classification. If the AI wasn’t confident, the system rerouted that transcript to a review queue. That single design element became key to building trust.

I introduced design system standards for EddyAI, creating reusable components for future products across the EdVisorly platform — consistent brand, consistent visual rhythm, and a cohesive design language.

Results & Institutional Impact

Once implemented, the results were dramatic.

At Stony Brook University, EddyAI automated more than 60% of transcript processing and eliminated duplicate records. Their admissions staff went from manually extracting data to spending their time evaluating students:

"We can now bypass the data collection part of the application reading process and focus on reading a student for the student."— Stony Brook University Admissions (from case study)

At Texas Tech University, transcript processing dropped from 2–5 minutes to about 30 seconds, and the system calculated multiple GPA versions automatically — something the institution had never been able to do manually.

EddyAI didn’t just save time; it shifted how teams worked.

Instead of wrestling with administrative tasks, teams could finally focus on student engagement, yield strategies, equity, and holistic review — the work that actually drives outcomes.

Reflection

This project was a milestone in my career because it exemplified what I love most about product design:

• Solving complex, high-stakes problems

• Designing for systems with real operational constraints

• Bridging technical architecture (AI/ML) with intuitive human experiences

The success of EddyAI wasn’t just the automation. It was the adoption — the moment institutional teams trusted the product because the interface made the AI’s decisions clear and auditable.

This work reinforced one of my core beliefs:

AI doesn't replace expertise. Great design gives humans superpowers.