Kelvin Chiu.

Portfolio

About

Kelvin Chiu.

Portfolio

About

Kelvin Chiu.

Portfolio

About

Kelvin Chiu.

Kelvin Chiu.

Portfolio

About

Short version

Full case study

Short version

Full case study

Conversational AI and AI-summaries for mental health patients

Role

Research, UX/UI Design, Product Strategy, Prompt Engineering, Testing Conversational AI

Senior Product Designer,

Mobile and Web

Team

Head of Product, Head of Design, Head of AI, CTO, Chief Clinical Officer, Senior PM, Front and backend Developers

Lack of engagement between therapy sessions

In early 2024, OpenAI reached a significant milestone with its launch of GPT-4o, and HelloSelf—a teletherapy platform connecting patients with licensed therapists—saw an opportunity: AI could finally play a role in making therapy more scalable and cost-effective. We were excited to build our first AI product to support patient progress.

Business Objective

Launch AI features fast and safely to build foundations for more scalable and cost-efficient therapy

User Problem

Patients often forget what's discussed, making it hard to maintain momentum between sessions and after therapy

Won £7M research grant with strong adoption

£7M grant to validate AI in therapy

Won a £7M Wellcome Trust grant to run the UK’s first clinical trial of an AI-supported therapy pathway—paving the way for the first clinically validated AI tool in mental health.

£7M grant to validate AI in therapy

Won a £7M Wellcome Trust grant to run the UK’s first clinical trial of an AI-supported therapy pathway—paving the way for the first clinically validated AI tool in mental health.

£7M grant to validate AI in therapy

Won a £7M Wellcome Trust grant to run the UK’s first clinical trial of an AI-supported therapy pathway—paving the way for the first clinically validated AI tool in mental health.

£7M grant to validate AI in therapy

Won a £7M Wellcome Trust grant to run the UK’s first clinical trial of an AI-supported therapy pathway—paving the way for the first clinically validated AI tool in mental health.

£7M grant to validate AI in therapy

Won a £7M Wellcome Trust grant to run the UK’s first clinical trial of an AI-supported therapy pathway—paving the way for the first clinically validated AI tool in mental health.

30% adoption within 3 months

AI summaries reached 30% adoption within 3 months. When conversational AI launched later, 1 in 3 summary users engaged with it in its first month of launch

30% adoption within 3 months

AI summaries reached 30% adoption within 3 months. When conversational AI launched later, 1 in 3 summary users engaged with it in its first month of launch

30% adoption within 3 months

AI summaries reached 30% adoption within 3 months. When conversational AI launched later, 1 in 3 summary users engaged with it in its first month of launch

30% adoption within 3 months

AI summaries reached 30% adoption within 3 months. When conversational AI launched later, 1 in 3 summary users engaged with it in its first month of launch

30% adoption within 3 months

AI summaries reached 30% adoption within 3 months. When conversational AI launched later, 1 in 3 summary users engaged with it in its first month of launch

4.3/5 user satisfaction rating

Patients rated AI summaries an average of 4.3 out of 5, and conversational check-ins 4.0—highlighting strong perceived value and user trust in a high-sensitivity space.

4.3/5 user satisfaction rating

Patients rated AI summaries an average of 4.3 out of 5, and conversational check-ins 4.0—highlighting strong perceived value and user trust in a high-sensitivity space.

4.3/5 user satisfaction rating

Patients rated AI summaries an average of 4.3 out of 5, and conversational check-ins 4.0—highlighting strong perceived value and user trust in a high-sensitivity space.

4.3/5 user satisfaction rating

Patients rated AI summaries an average of 4.3 out of 5, and conversational check-ins 4.0—highlighting strong perceived value and user trust in a high-sensitivity space.

4.3/5 user satisfaction rating

Patients rated AI summaries an average of 4.3 out of 5, and conversational check-ins 4.0—highlighting strong perceived value and user trust in a high-sensitivity space.

It’s like Gemini or Siri—but for therapy. It gives me something smart I can turn to that actually knows what I’ve been through.

- Patient

It’s a helpful reflection point. I can revisit what I was feeling, see what’s changed, and think about how I want to keep improving.

- Patient

0-to-1 Designs on Mobile and Web

Desired outcome:

Introduce AI to patients at a moment that feels natural, low-effort, and genuinely useful so that it becomes a part of their therapy.

New AI-Ready Call Station

I redesigned the UX/UI of the call station so patients could consent-to/activate AI features alongside their therapist—making the experience clear, collaborative, and non-disruptive.

AI-generated Summary

AI-generated Summary

I designed the UX/UI for reviewing session summaries to support memory recall—making the experience simple and easy to return to.

I designed the UX/UI for reviewing session summaries to support memory recall—making the experience simple and easy to return to.

Conversational AI

I designed the UX/UI for how patients chat with the conversational AI to reflect on their sessions. I also added a feedback mechanism to help the team improve model responses over time—making the experience feel human and safe to use.

LLM Evaluation

LLM Evaluation

Working with Head of AI and Chief Clinical Officer, I helped test and evaluate the quality of AI-conversations based on different synthetic personas I generated.

Working with Head of AI and Chief Clinical Officer, I helped test and evaluate the quality of AI-conversations based on different synthetic personas I generated.