REIMAGINING LEARNING SOCIAL EMOTIONAL SKILLS

Social Cue is an AI social coach designed to transform social uncertainty into confidence-building platform that helps K–12 learners—especially neurodiverse students—build real-world social and emotional skills. Using AI-driven emotional recognition, adaptive feedback, and voice-based conversation practice, it guides users through everyday interactions like making friends, managing conflict, and expressing emotions. The long-term vision includes optional wearable insights to deepen emotional awareness and real-time support.

Social Cue transforms social learning into an interactive, inclusive, and empowering experience that bridges the gap between classroom instruction and authentic human connection.

Real-time coaching for real-world connection.

THE PROBLEM

A Design Failure, Not a Deficit.

Zach, a seventh grader, stands at the edge of the cafeteria, tray in hand, scanning for somewhere to sit. He knows the rules. He learned them Thursday. In this moment, flooded with noise and uncertainty, he can't access them. Maya rehearsed the conversation three times before school. By the time she reached her locker, the script was gone. Jordan can tell you the five steps to starting a conversation. He can't tell you why it never feels like him when he follows them.Lily passed the social skills assessment. She still eats lunch alone.

These aren't outliers. This is the daily reality for millions of neurodivergent students. Social-emotional learning shows lasting benefits, yet only 5 of 166 studied interventions were tested for students with disabilities. The programs that do exist teach kids to perform neurotypical behavior rather than understand what's happening around them. Skills practiced in structured settings don't transfer to chaotic hallways and cafeterias. Compliance metrics track whether students perform expected behaviors, not whether they understand social dynamics or feel like themselves.

The harm goes deeper than ineffectiveness. Research links social camouflaging to anxiety, depression, identity confusion, and suicidality. Masking teaches kids to constantly judge their own social exchanges, encoding the message that their authentic way of connecting is the problem. Meanwhile, self-advocacy and self-determination predict better outcomes across education, employment, and quality of life.

The design challenge: how do you give kids social comprehension and agency without encoding the message that neurotypical communication is the "correct" standard?

DESIGN PROCESS

We Didn't Start With a Solution. We Started With Listening.

I began with no assumptions and no predetermined solutions. From April through June 2025, I conducted semi-structured interviews with fifteen stakeholders and spent over 40 hours observing students in naturalistic settings at iLEAD Lancaster, a public charter school in Southern California.

I tracked interactions in classrooms, during lunch transitions, and in hallway moments where social anxiety quietly peaks. I documented patterns, contradictions, and moments that didn't fit into neat categories.

Working alongside Katelyn Kaufman, CCC-SLP at Ohana Student Services, one insight fundamentally reframed everything: students weren't struggling because they lacked skills. They were struggling because the support they received was designed to make them perform neurotypicality, not to help them understand what was happening or decide how they wanted to respond.

This became my north star.

Teachers described spending 2-3 hours weekly managing emotional escalations with no real-time tools. One special education teacher, Mrs. Park, documented seventeen separate incidents in a single week where students shut down during unstructured social time. During observations, I witnessed nine instances where students stood at the periphery of group activities wanting to participate but unable to initiate. Six students physically isolated themselves rather than risk saying the "wrong thing."

Parents shared quiet frustrations about gaps no program could fill:

This ethnography didn't just inform my process. It shaped the entire direction of the solution.

"Many of my students have disabilities such as autism or speech-language differences, and social skills are one of their biggest needs. They rely on support with boundaries, conversation skills, and emotional regulation every single day. Kids don't need to be fixed. They just need tools to navigate a world that doesn't make sense to them." (Katelyn Kaufman, CCC-SLP)

"It's very difficult to watch our son Zach misread a peer's disinterest as enthusiasm, leading to painful rejection." (Thomas & Rachel Fielder)

15 participants across 3 stakeholder groups
7 students ages 7-16 (5 with IEPs for autism, ADHD, or anxiety)
3 teachers, 1 speech-language pathologist, 1 occupational therapist, 1 school counselor
2 families of neurodiverse learners
Classroom observations spanning arrival routines, collaborative learning, lunch transitions, and dismissals
Individual therapy sessions and small-group interventions

RESEARCH PARAMETERS

THE HIDDEN PATTERN

Students Weren't Resisting Support, They Were Resisting Exposure

My first hypothesis was completely wrong. I entered research believing students needed better instruction in social skills. Within three interviews, that assumption collapsed. Students kept expressing what seemed like contradictory needs:

"I need help." <-------> "Getting help is embarrassing."

Then one student reframed everything: "I don't mind help. I just don't want anyone else to see me getting help."

The issue wasn't intervention. It was audience. Students wanted guidance that felt invisible to peers but powerful to them.

What Students Told Me:

"I want to know how to make friends. Like, actually make them. Not just be in the same room."

(11-year-old with ASD)

"I wish someone could tell me what I'm doing wrong without everyone seeing. It's embarrassing when the teacher has to explain things to me in front of everyone." (14-year-old with social anxiety)

"It feels like everyone else got a rulebook except me. They just know stuff I have to figure out." (13-year-old with ASD)

Students weren't asking for therapy or traditional social skills classes. They were asking for private, judgment-free space to understand what was happening around them.

THE PIVOT

I Was Building The Wrong Thing.

Then something stopped me. My early prototypes reflected a belief I didn't know I held: that the goal was helping neurodivergent kids pass as neurotypical. My prompts were coaching scripts for assimilation. One student told me, "It feels like being told everything I do wrong." Another said, "You want me to act like everyone else."

She was right. I had designed a tool to teach masking, the exhausting performance of normalcy that neurodivergent people know intimately. I was building the thing the research told me causes harm.

Studies link social camouflaging to anxiety, depression, identity confusion, and suicidality. Masking teaches kids to constantly judge and dissect their own social exchanges, encoding the message that their authentic way of connecting is the problem. Meanwhile, self-advocacy and self-determination predict better outcomes across education, employment, relationships, and quality of life. The double empathy problem, research showing that communication breakdowns between neurotypical and neurodivergent people are mutual, suggests that one-sided interventions address half the equation.

I pivoted completely.

REFRAMING THE DESIGN CHALLENGE

The question evolved from:

How might we provide better SEL support to neurodiverse students?

How might we help students understand social dynamics so they can make their own choices?

How do you give kids social comprehension and agency without encoding the message that neurotypical communication is the "correct" standard?

WHAT THE RESEARCH REVEALED

Effective support must....

BUILD COMPREHENSION, NOT COMPLIANCE

The goal isn't teaching students how to act normal. It's making social information legible so they can understand what's happening and decide how they want to respond. Their way of connecting isn't wrong. It's different. The tool should offer options, not prescriptions.

FEEL PRIVATE, NOT PUBLIC

Traditional interventions inadvertently highlight differences. Students become hyper-aware of being watched, corrected, or singled out. True learning requires psychological safety, and psychological safety requires privacy.

"I wish someone could tell me what I'm doing wrong without everyone seeing."

PRESERVE AGENCY AND IDENTITY

Students don't want to be told what to do. They want to understand what's happening so they can choose for themselves. Support should offer multiple response options reflecting different values: direct, warm, protective, curious, self-advocating. The learner decides what fits.

"I want to practice where it's okay to mess up. Where nobody's going to laugh or think I'm weird."

EXIST IN THE MOMENT, NOT AFTER THE FACT

Social challenges happen in real time. Feedback delivered hours or days later doesn't create the learning that changes behavior.

"I can teach a student how to join a conversation in our session. But when the actual moment happens at lunch, I'm not there. And by the time they come back to see me, they've already internalized whatever happened as another failure."

FROM INSIGHTS TO IDEAS

The Ideation Process

Once I understood the real challenges, I moved into ideation without immediate filtering. I explored multiple directions:

The concept that aligned best with student needs was a hybrid approach: AI-powered practice accessible through an app, with future integration of discrete wearable devices for real-world coaching.

This wasn't the flashiest idea. It was the idea most grounded in what the research revealed people actually needed.

Simple conversation guide tools students could reference during interactions
Peer buddy systems pairing students with trained mentors
VR-based practice environments for immersive social rehearsal
Wearable technology providing physiological feedback
AI-powered coaching accessible through familiar devices for private practice

“The ideation process taught me something important about design: creativity isn't about generating as many ideas as possible. It's about deeply understanding the problem and then letting that understanding guide you toward solutions that actually make sense for the people you're serving.”

- Kelsey Evenson (SocialCue Founder)

BUILDING & TESTING: FROM CONCEPT TO FUNCTIONAL PROTOTYPE

I started with deliberately simple prototypes. No polished interfaces, just the basic structure of interaction I needed to test. Using a "Wizard of Oz" approach, I manually triggered AI responses to test emotional tone, pacing, and conversational structure without getting bogged down in technical implementation.

After refining the basic structure, I built a functional prototype integrating:

Real conversation practice through voice interaction
Dynamic AI responses that adapt to student input and performance
Grade-adaptive content that adjusts language complexity and scenario difficulty
Progress tracking with specific, encouraging feedback

TESTING WITH REAL STUDENTS

EACH 20-MINUTE TESTING SESSION REVEALED SPECIFIC REFINEMENTS:

ITERATION 1 TO 2: EMOTIONAL TONE

Early versions felt too clinical. I rewrote the AI's voice to sound warmer, more like a supportive older sibling than a therapist.

ITERATION 3 TO 4: FEEDBACK SPECIFICITY

Generic praise wasn't landing. I designed the system to identify and name specific behaviors: maintaining eye contact, using someone's name, asking follow-up questions.

ITERATION 2 TO 3: SCENARIO VARIETY

Students got bored with repetitive situations. I expanded scenarios from simple greetings to complex social navigation:

"how to join a conversation that's already happening" or "how to respond when someone seems upset with you."

ITERATION 4 TO CURRENT: AVATAR INTERGRATION

Students needed visual representation. I developed an avatar that models expressions, tone, gestures, and body language contextually while simultaneously detecting the learner's emotional cues to adapt conversations in real-time.

WHAT HUMAN-CENTERED DESIGN REVEALED: FIVE CORE PRINCIPLES

These principles didn't come from theory. They emerged from listening to students, observing real moments of struggle, and testing what made support feel empowering rather than uncomfortable.

PRINCIPLE 1: SUPPORT SHOULD FEEL INVISIBLE, NO EXPOSING

Students wanted guidance without being singled out. Social Cue provides private, judgment-free practice that doesn't call attention to the learner.

PRINCIPLE 2: REAL-TIME SUPPORT MATTERS MORE THAN PERFECT SUPPORT

Social challenges happen in hallways, lunchrooms, and transitions, not in scheduled sessions. By the time feedback arrives later, the moment has passed and the student has already internalized the experience as failure.

PRINCIPLE 3: SOCIAL-EMOTIONAL SKILLS ARE PHYSICAL AS MUCH AS EMOTIONAL

I observed students' bodies signaling stress before they spoke: clenched fists, shallow breathing, tense shoulders. Social Cue's AI analyzes tone, pace, and vocal expression to detect anxiety or confidence, building students' awareness of the mind-body connection.

PRINCIPLE 4: TEACHERS NEED PATTERNS, NOT MORE TASKS

Teachers don't need another system to monitor. They need to see social-emotional patterns without extra work. Future teacher dashboards will focus on high-level trends: who is struggling, when, and why.

PRINCIPLE 5: FEEDBACK SHOULD PROTECT DIGNITY

Students responded best to warm, encouraging language that assumed they were capable. Social Cue normalizes struggle, celebrates effort, and frames mistakes as information rather than failure. The goal is always to leave the learner feeling more confident than when they started.

These principles now guide every iteration of Social Cue, from current features to future expansions including wearable integration, real-world coaching through earbuds, and multiplayer practice environments. The research revealed not just what to build, but why it matters and how to build it with dignity at the center.