From “Her” to Here: How Sten is Bringing Samantha to Life
Inspired by the emotional depth of the AI assistant Samantha from the movie Her, Sten is an AI-powered journaling app designed to foster self-reflection and emotional well-being. By leveraging multi-modal emotion recognition, Sten creates a personalized journaling experience that understands and responds empathetically to users’ emotions—bringing a new level of emotional intelligence to digital interactions.
Prototype linkMost journaling apps simply store thoughts without truly engaging users. But with mental health concerns growing, people need more than just a digital diary—they need a companion that listens and responds.
That’s where Sten comes in. Using emotion AI, voice recognition, and real-time analysis, Sten transforms journaling into an interactive, therapeutic experience. By detecting emotions through text, voice, facial expressions, and body language, Sten doesn’t just record feelings—it helps users understand and process them, just like a thoughtful conversation.
My Approach
To build an empathetic AI-powered journaling experience, I took a user-centered approach, blending design thinking, psychology, and AI-driven emotion recognition to create Sten.
1. Understanding Users & Their Needs
I spoke with users through interviews, surveys, and empathy mapping to explore how they journal and what they need. The key takeaways? They wanted emotional feedback, struggled with consistency, and worried about privacy when AI analyzed their emotions.
2. Defining the Problem & AI-Driven Solutions
From research, I identified three major challenges:
- Journaling apps feel passive—users want more engagement.
- People abandon journaling—lack of motivation.
- Privacy concerns – users need control over emotional data.
The solution? Sten would detect emotions, provide real-time insights, and offer AI-driven prompts to encourage reflection.
3. Designing an Emotionally Intelligent AI
I developed a system where AI understands and responds to emotions:
- Emotion Detection → NLP for text, CNNs for voice, VGG16 for facial recognition.
- AI-Powered Prompts → Personalized based on emotional state.
- User-Friendly Design → Calm, intuitive, and engaging UI.
4. Testing & Refining the Experience
I tested Sten with real users, using SUS surveys & product reaction cards to gather feedback. This led to:
- More user control → Opt-in for AI-driven emotional tracking.
- Better journal reflection UI → Easier navigation.
- Stronger privacy settings → More transparency & trust.
Results
Sten successfully bridged the gap between AI and emotional well-being, transforming journaling from a passive habit into an interactive, self-reflective experience.
Increased Emotional Awareness – 83% of users found AI-driven reflections helpful for identifying emotions, leading to deeper self-awareness.- Higher Engagement & Retention – AI-generated prompts and emotional insights increased journaling consistency by 40%.
- Improved Usability & Privacy – User testing led to key refinements, including better UI for emotion tracking and an opt-in privacy model for AI-driven emotional analysis.
- Bridging AI & Empathy – By integrating multi-modal emotion recognition, Sten achieved human-like emotional intelligence, making it the closest digital companion to the AI from Her.
Integrated multi-modal emotion recognition to deliver human-like, empathetic AI interactions.
Real-time notifications reduce last-minute cancellations and travel uncertainty.
View Figma Prototype here
40%
increase in
engagement
83%
users found AI reflectionsvery helpful
Future Plans
-
Expanding Emotional AI Capabilities – Enhancing real-time AI feedback with more nuanced emotion detection and personalized journaling prompts.
-
Stronger Privacy Controls – Offering customized AI settings so users can control emotion tracking and data storage preferences.
-
Integration with Well-being Practices – Adding guided mindfulness exercises and mood analytics to further support users' mental health.
-
Conversational AI Evolution – Improving Sten’s AI interactions to provide even more natural, empathetic responses in journal reflections.
