Tanya Sheremeta
Interactive Designer, Illustrator, and Animator

CURIO | VUI for Kids

CURIO is an engaging, individualized conversational assistant that helps children with ASD reach their education goals. 


Compatible with all major learning curriculums, CURIO is a voice interface that increases one on one interaction, instructional scaffolding, progress monitoring, and student engagement— making it perfect for children in special education programs.

banner_banner for web.jpg

vui, ux/ui, motion


My partners and I wanted to explore developing a product that focused on educational creativity and achievement. We were concerned that most STEM toys were economically out of reach for many families. We narrowed our focus to finding opportunities where a conversational platform could assist children in the classroom. 

To draw inspiration for concepts, I began by performing a literature review of scholarly articles and publications. I focused on the current use of assistive AI in therapeutic settings. This led me to case studies where the potential of assistive AI with kids on the autistic spectrum were being explored. I saw an opportunity and proposed creating a conversational AI platform that would close the achievement gap for at risk students in classroom settings.

read full research brief here.


challenge: Student project to design an educational robot for children with a conversational interface.

my role:
literature review: medical, robotics, convo AI
research (bibliography)
product thinking
data visualizations
illustration of child and robot
motion graphics: logo animation, tagline animation, robot interactions, child animations, magic

Sean O’Leary
Lexi Korobkin

tools: illustrator, photoshop, procreate, after effects

timeline: 6 weeks


Approximately 6.7 million students receive special education services. This comprises 13% of total public school enrollment.



students: Some special education students are not reaching their potential. A child with autism spectrum disorder might have trouble communicating verbally, paying attention to others, or controlling their stress and anxiety. When students are trying to learn at a pace they can’t keep up or participate in it is easy for them to fall behind. These difficulties can affect the child’s social development and their success in school. 

teachers: Educational professionals work with multiple students with varying needs and on multiple objectives using different interventions to achieve each student’s unique goals; what works well for one child might lead to discomfort or low metrics for another.  

parents: Parents are concerned with stunting social skills by substituting human interaction with robots. They are also concerned with privacy issues and excessive screen time.


Struggling students may think they are inhibiting the learning process for the whole group, which only increases their anxiety. These stresses play a significant role in educational success.


Competitors have focused on the communicative and social limitations of children instead of core educational competencies. Most of these robotics are not autonomous, but rely on aides to operate them “wizard of oz” style.


“The key to solving the student achievement gap is implementing evidence-based practice with fidelity.”

Stevan J. Kukic, Ph.D., Director, School Transformation
National Center for Learning Disabilities


understanding the user:

Based on my research, we were able to validate the concept and also identify user problems to aid in thinking about product development. My teammate conducted user interviews with a special education specialist and a speech therapist in the greater Seattle area.  From the transcripts we were able to identify key insights into how special educators work with kids and define pain points for both the teachers and the children. From here, we were able to map key user stories to guide us in developing our core features. 

see whiteboarding the MoSCoW and the child journey map


For Meghan’s kids, the fear of thinking they’re different is really impactful. They do best when classmates know that each student has a different plan.

journey map-05-05.png


My teammates were very affected by the interview of the speech therapist. She described using current technologies like Smartboards in the classroom and couldn’t see the use of a conversational tool.  My teammates thought developing a teacher and separate student facing app was the most logical step in our product development.

I thought pivoting away from working solely on the conversational AI was a missed opportunity, but I recognized that augmenting voice platforms with screen interfaces can help students with speech problems overcome communication barriers. Some children may need visual aids in order to achieve their potential.



While my partners worked on the design system for the screens, I storyboarded our persuader video and worked on the logo animation, tagline, and introduction to the robot and child.

In the video we describe how Curio adapts educational activities to the needs of children and further tailors curriculum to the specialized objectives of educators. Interaction, including activity responses, instructional goals, and mood metering are tracked in real time, recorded securely in the cloud, and assist teachers and aides with engaging students during lessons. Teachers can quickly set up instructional profiles for each student, link IEP goals to curriculum, and adjust prompts, responses, and desired feedback. The goal is to use voice technology to improve educational access. 

read full features here.


The goal is to utilize technology to create an educational assistant that increases access to support for kids and allows teachers to refine progress expectations and strategies immediately.

Comp 2_1.gif

designing the robot:

The human face, with its vast range of complex expression, can be stressful. A robot is predictable and has bottomless patience. A robot can provide correction without judgment and in doing so may provide relief from the pressure to understand emotional cues. A robot is always polite, which can encourage a child to persevere in the difficult work of expanding skills.

While my partners designed the companion app, I designed Curio with “minimal” emotional expressiveness in order to be uncomplicated and easy to interpret. The robot is simple in both shape and color while still holding to an abstracted humanoid form.  Curio is designed to sit tabletop to reinforce that the bot is a tool not a toy.

Feedback from teachers recommended removing arms and arm motions.  They thought the kids would not take it seriously or think that it was funny, but not in a constructive way.


the conversation:

CURIO was designed to model empathetic behaviors through speech through active listening, reflection, and by adopting a nonjudgmental warm tone. A human can’t repeat an instruction in exactly the same way with the same tone of voice, a robot can repeat ad infinitum.  Robots can provide a predictable and consistent environment to practice skills.

The bot uses simple techniques like maintaining conversational initiative by asking questions, retaining a sense of coherence by referring to the child’s previous utterance, and using simple pattern-matching rules to generate system responses. Using automatic speech recognition and spoken language software, the robot can prompt the child by keeping track of what information is required and asking questions accordingly.

I illustrated a conversational flow to explore how a student would engage with CURIO to perform a specific task, and how possible errors would be handled.


Studies have shown that practice conversation translates into more facility with peers: a back and forth interaction with logical trajectory can increase a child’s success in school.

Comp 1.gif

reflection + moving forward:

We understand that children build relationships with conversation agents the same way that they build relationships with people. Children also attribute intent and emotion to objects that they engage with socially and psychologically. We anticipate that kids would playfully test our device’s limits by repeatedly asking the same question to see if the bot would change its answer. I would like to test and observe this probing behavior by setting up test scenarios with kids using SayWizard. 

The brief I wrote was far too lengthy to be a guiding document. I think we would have benefited from crafting an intent statement that clarified our project’s trajectory at each stage. In the future, I will break down my research into more palatable sections and codify our concepts as we go.