Case study · Now a live product at andilabs.education
A Master's HCI capstone for Logan & Friends that gave teachers real-time, AI-powered coaching instead of one observation a year. It has since grown into ANDI — a live coaching platform for teachers, coaches, principals and district leaders.

00
The brief
Dr. Jocelyn Logan-Friend had been a classroom teacher, an instructional coach, a coach to school leaders, and a federal education specialist. She had watched the same gap open up at every level — and she brought it to our capstone team at Indiana University.
The ask was deliberately open-ended: teachers need continuous, personalised coaching, and almost no school can afford to staff it. Design something that could. Over eight months, four of us researched, prototyped and tested what that might actually look like in a real classroom.

An eight-month industry capstone pairing Logan & Friends with the Luddy School of Informatics, Computing and Engineering.
Chapter 01
Every teacher is told to improve. Almost none of them are shown how more than once a year.
01
The problem
Formal observations happen once or twice a year. In between, teachers are on their own — and the coaches who could help are stretched across too many classrooms to be anywhere near enough of them.
Logan & Friends brought my capstone team a gap they saw across schools: districts pay real money for professional development they can't prove is working, and instructional coaches who want to be more present for their teachers physically can't be. We set out to design something that could sit in the room with a teacher every day — not once a semester.
The gap the project set out to close: annual evaluation on one timeline, daily practice on another.
1 in 4
Teachers who say PD actually improves their teaching
$18K
Average PD spend per teacher, per year — with little measurable impact
50:1
Average teacher-to-coach ratio in U.S. districts
Figures ANDI Labs cites today — the same gap this project set out to close.
Chapter 02
Before designing anything, we needed to know exactly where the feedback broke — and for whom.
02
Research & discovery
This started as my Master of HCI capstone at Indiana University — an 8-month engagement with a team of four, working directly with Logan & Friends founder Dr. Jocelyn Logan-Friend.
We ran interviews, surveys, and workshops with classroom teachers and instructional coaches to understand where feedback actually broke down — and built two personas to keep the team honest about who we were designing for.
PERSONA · 01
Wants to grow but only hears from an evaluator once or twice a year. Doesn't want to be judged — wants a low-stakes way to see what's actually happening in her own classroom and try something different next class.
PERSONA · 02
Responsible for far more teachers than he can visit regularly. Needs to know where to spend his limited time — and a way to stay looped in on teachers he can't get to in person every week.
01
By the time a formal observation report arrives, the lesson it describes is long over. Teachers needed feedback close enough to the moment to actually act on it.
02
Coaches wanted to be everywhere and couldn't be. The fix wasn't more coaches — it was giving the ones who exist enough visibility that their limited time goes to the teachers who need it most.
03
Teachers were far more open to hard feedback when they could sit with it on their own terms before deciding whether to loop in a human coach at all. Privacy wasn't a nice-to-have; it was what made honesty possible.
04
The moment we framed the AI coach as a replacement for human coaching, trust dropped sharply. Framed as a bridge that helps a teacher arrive at a coaching conversation better prepared, it held.
Chapter 03
Four features, one job — put useful feedback in a teacher's hands while the lesson is still fresh.
03
The design
Every feature ties back to the same job: get a teacher useful, specific feedback close to the moment it happened — and give coaches a way to be present without being everywhere at once.
FEATURE · 01
A teacher records a lesson and gets back a timestamped recording they can scrub, AI-generated feedback with actionable insights, color-coded domains for quick scanning, and suggested resources for what to try next.

%201.jpg)
FEATURE · 02
Teachers can talk through their feedback with the AI coach first, then choose to tag a human coach for a second perspective when they want one. Every report is saved to a personal hub they can revisit.
FEATURE · 03
An AI assistant is available mid-recording — a teacher can ask for an activity idea or a different way to explain a concept, and get an answer without breaking stride.


FEATURE · 04
A resource hub of videos, articles, and podcasts, filterable by domain and format — so feedback points to something concrete a teacher can go learn from, not just a score.
Chapter 04
Same playbook, different game — close enough to compare, distinct enough to matter.
04
The prototype
“Same playbook, different game — close enough to compare, distinct enough to matter.”
We designed the teacher and coach experiences as one visual system so a coach could recognize a teacher's dashboard at a glance, then gave each role only what it needed.
.png)
The teacher's dashboard and Learning Hub — sessions this month, tagged reports, performance across domains, and upcoming sessions in one view.
TEACHER
Sessions this month, tagged reports awaiting review, performance trends across domains over time, and an upcoming-sessions calendar — everything a teacher needs to see their own growth in one place.
COACH
A near-identical layout, minus the record and upload tools a coach doesn't need, plus a roster of assigned teachers, their tagged reports, and the ability to open any teacher's profile and respond to what's flagged.
INTERACTIVE
Eleven flows from the working Figma file — everything from first sign-up through to the coach's side of the product. Pick one and click through it exactly as a teacher would.
05
Testing & iteration
We ran usability tests using think-aloud sessions to catch what the wireframes couldn't tell us — where people hesitated, misread a label, or reached for the wrong icon.
01
Both started as plain text buttons sitting side by side. Testers hesitated on each one until we gave them distinct icons, and then distinct colour — a camera for Record, an upload arrow for Upload.
02
The sidebar went through two rounds of simplification before landing on icons paired with plain labels — Home, Reports, Learn, Saved, Calendar, Chat. Testers paused on anything less literal.
03
A dedicated domain-by-domain breakdown — an overall proficiency gauge plus per-domain scores — was a gap testers named directly and unprompted. We scoped it and it shipped in the next round of design.
04
Asking people what they thought of the design mostly produced politeness. Watching them narrate a real task produced the hesitations that actually drove every change above.
Chapter 05
The capstone ended. The product didn't.
06
Where it is now
After graduation, I was one of two designers from the original team asked to keep going with Logan & Friends on the AI coach — five more months of usability testing, interviews, and workshops, expanded to school administrators and district-level professionals. Today it's live as ANDI.
.jpeg)
At the centre of it is the CIQ — the Classroom Impact Quotient, ANDI's proprietary scoring framework that turns a recorded lesson into one clear, actionable number for how effectively it drove student engagement and learning.
One product now serves four audiences: teachers get a private development loop, coaches get visibility across everyone they support, principals get coaching at every seat without adding headcount, and district leaders finally get a system-wide view of what their PD spend is changing.
“The data existed. The research existed. The technology existed. What didn't exist was a way to bring all of it into the classroom — in real time, without adding to the burden teachers were already carrying.”
— Dr. Jocelyn Logan-Friend, Founder & CEO, ANDI Labs
07
Reflection
Collaborating with engineers, founders and educators taught me how differently each reads the same problem. The work was agile and non-linear — research and design rarely moved in a straight line, and staying adaptable when the plan changed mattered more than defending the plan.
I led presentations, workshops and interviews — and got past a real fear of public speaking doing it. I also navigated genuinely difficult team dynamics, which taught me more about conflict resolution and empathy than any coursework did.
“Apoorva has strong organizational and leadership skills. As a leader on a team of four, she excels at managing tasks, asking key questions, and keeping the team motivated.”
— Professor Julie Stella, Capstone Guide
“When faced with the technical hurdle of prioritizing features for our project's target users, Apoorva remained undaunted. Her design thinking skills led us to a breakthrough solution.”
— Pavan Suresh, Product Designer
This is the project I point to when someone asks whether student design work can hold up outside the classroom. See it live, or head back to see what else I've been building.