New Zealand AI Tool Spots Depression Before It Escalates

When Kunal Gupta felt depressed in his mid-20s, he had nowhere to turn. Growing up in India, talking about mental health felt impossible, and when he tried, his family told him to just put down his phone.

Now the 36-year-old researcher at the University of Auckland is building technology to help others avoid that isolation. His team developed Tōku Hoa (My Friend), an AI mental health companion that could spot trouble before it spirals.

The system connects to your smartwatch and phone, tracking signals like heart rate, sleep patterns, and activity levels. Custom algorithms look for patterns that suggest someone might be struggling emotionally.

When it notices something concerning, the AI sends a simple message: "Hey buddy, we noticed you're feeling stressed. Do you want to check in?" Users can then talk through what happened via text, voice, or video using techniques borrowed from cognitive behavioral therapy.

Early trials show the system can identify emotional states with 80 to 85 percent accuracy. The team is now testing it specifically with New Zealand populations, focusing first on young adults aged 18 to 25.

Gupta emphasizes this isn't meant to replace therapists. "It's kind of like a bridge between when the person starts experiencing mental health challenges and when they actually reach a therapist," he said.

The team plans to work closely with Māori and Pasifika communities, particularly young men who face disproportionately high mental health risks. New Zealand's long wait times for mental health services make tools like this potentially valuable for people stuck in limbo.

Why This Inspires

What makes this project special is its origin story. Gupta built this because he remembers desperately wanting "someone who could just listen" without judgment. His personal experience with stigma shaped every feature, from the gentle check-ins to the focus on reflection rather than diagnosis.

Dr. Kieran Dang from digital health platform Mosh cautioned that AI health tools need robust testing and clear safeguards, especially for serious issues like mental health. "If there's ever any doubt, people should always play it safe and seek medical advice from a real medical professional," he said.

The prototype is still being tested in laboratory settings, but Gupta's vision is clear: create the supportive presence he wished he'd had when he needed it most. Sometimes the best innovations come from people who've lived through the problem they're trying to solve.