No time for the details? Click through the summary slider below to understand my key contributions and takeaways.
Psyk AI is a multimodal AI conversational agent platform project, aimed at student therapists. Our team designed this concept for an entrepreneurship competition, and have since undertaken it as a serious product.
The idea for this product germinated from a very personal problem: my team members had consistent issues finding therapists with the training needed to treat POC-specific issues.
Psk AI addresses this problem by focussing on improving training outcomes for student therapists, by giving students a safe space to train on a wide library of interactive scenarios, gain closer-to-real-world competency, and avoid harming POC.
Our experience has taught us that our product offering needs stronger product-market fit. We still haven't uncovered the 'hair on fire' problem.
Our next goal is developing our MVP, and presenting that to early adoption clients.
Psyk AI is an AI-powered co-pilot trained on synthetic data, PsykAI provides psychology students with a safe virtual environment to practice therapy sessions with conversational agents,” Naveed said. “This unique approach allows students to develop their cultural competence, hone decision-making skills and refine therapeutic techniques without risking harm to real clients or compromising confidentiality.”
The Gen AI Challenge was a competition designed to foster innovation, problem-solving skills, and collaboration among students at the University of Michigan.
Teams were required to come up with product proposals that used Gen AI in strongly needed, socially positive, and business-viable.
The competition provided an opportunity for participants like our team to showcase their knowledge and expertise by utilizing artificial intelligence tools (AI) while tackling real-world challenges. Through this competition, we honed our entrepreneurial mindsets, experimented with cutting-edge AI technologies, and developed interdisciplinary collaboration.
This was the team's first competition, so we were all excited to knock it out of the park! But first, we had to plan around the competition phases.
Concept Exploration
Nov:
Our team brainstormed concepts, interviewed specialists, analyzed solutions according to business and technical viability, and narrowed our concept down to one.
8 Finalists Chosen
Dec:
8 finalists were chosen across the entire university, and would begin working on final presentations in Jan.
The Final Presentation & Concept
Jan:
We interviewed Therapy, Social work, Healthcare, & Gen Ai specialists, to refine our presentation and prototype.
We presented, and earned 2nd place!
Our team members all had multidisciplinary backgrounds; from Anthropology, Psychology, and Non profit work, all the way to Project Management, Technology and Business consulting.
We evaluated our ideas according to models required, problem space, social impact, pricing models, competitiveness, and dev effort.
Our top choices were:
+ AI legal services and multilingual documentation for immigrants
+ Live CBT therapy session analytics
+ AI powered medical diagnostics chatbot
Our final concept was a copilot for therapists.
Our envisioned product was dubbed PsykAI: An AI-powered mental health training assistant that aimed to transform and enhance the learning experience for psychotherapy trainees.
The tool would leverage machine learning to simulate psychotherapy sessions, providing trainees with low-stakes yet realistic practice environments to build competency and expertise before working with actual clients.
We envisioned PsykAI to operate through a phased approach — beginning with AI assistance, progressing to human-AI collaboration, and culminating in autonomous AI-led interactions. By supplementing traditional supervised training with scalable and adaptable AI support, PsykAI would help minimize errors and potential harm to clients during the trial-and-error learning process.
It represents an innovative integration of AI into mental healthcare training to solve the problem of variability in therapist competency development.
Once we made it past the concept proposal stage, we began interviewing specialists at the University of Michigan's School of Social Work, Open AI, Red Cross, the Center for Entrepreneurship and more.
Through these discussions, we uncovered several key problems:
Therapists Don't Need Help; Students Do
Our tool was initially designed to augment therapists care, but they don't strongly need support.
But students do.
The Tech Wasn't There Yet
Speech to text, sentiment analysis, & text to speech processing is still too slow to keep up with regular conversation.
HIPPA, Insurance & the Law
Between figuring out how to comply with HIPPA, prevent insurance companies from using transcripts to deny care, and dealing with possible govt. subpoenas, there were so many hurdles involved around working with real client data.
After all our insights within the team, we decided to pivot towards a training tool for therapist in training.
After weeks of hard work (and 2 all-nighters), we presented our final product vision, MVP and business and technology roadmaps.
Click through the FIGMA link above to check out our final presentation!
I would be lying if I said we weren't disappointed in getting 2nd place. But that didn't change our hard work and belief in our solution.
However, there are several weaknesses that we have started working on:
We really believe in this idea. The concept, business projection, and team are strong, and we're looking to refine the product offering through competitions, mentorship, and ultimately funding.
So be sure to revisit this page in the future; there might be an update!
