AI for mental health: Reflexion in action

Transforming Well-Being With AI

Reflexion’s latest study—conducted with 4 200 university students over a three-week exam block—shows a 32 % reduction in reported stress levels after nightly 10-minute Reflection Sessions. By combining affective-computing models with cognitive re-framing prompts, the platform guides users from raw emotion to actionable insight in under five minutes.

“I opened Reflexion between back-to-back finals and felt my heart rate slow almost immediately. It’s like a guided conversation with myself.” — Jordyn K., senior at UCLA

Real-Time Emotional Insights

At the core of Reflexion is an on-device transformer that classifies sentiments, detects cognitive distortions, and proposes micro-prompts— all within 120 ms. Because the processing stays on the phone, users keep full control of their data while still benefiting from powerful analytics.

• Sentiment arcs visualize mood shifts across a day.
• Thought traps surface patterns like catastrophizing or black-and-white thinking.
• Synthesis snippets generate a one-sentence takeaway for quick reflection.

Clinically-Aligned Prompts

Reflexion’s prompt library is co-authored with licensed therapists and maps to evidence-based modalities such as CBT and ACT. Each suggestion is graded for readability (Flesch score ≥ 62 ) to ensure accessibility for users as young as 15.

“The AI’s re-framing prompt mirrored the advice I give my patients — impressive accuracy for a consumer app.” — Dr. Anika R., clinical psychologist

Looking Ahead

Upcoming releases will add heart-rate variability integration from popular wearables, enabling Reflexion to time check-ins precisely when physiology indicates rising stress. The team is also piloting a “Peer-Share”  mode, letting friends send supportive prompts—a social layer shown to boost adherence by 18 %.