CoachLens AI

CoachLens AI is an empathetic app that analyzes recordings of parent-child interactions to provide a professional-grade coaching session. It transforms everyday family moments into measurable growth.

Comment

Markdown

🍼 CoachLens AI

Transforming Parenting Theory into Daily Practice through Multimodal AI Coaching.

🌟 Inspiration

As new parents in 2025, we quickly realized that while there is no shortage of parenting books and videos, there is a massive gap between knowing a technique and executing it in the heat of the moment. We wanted to raise a child with high emotional intelligence and a positive mindset, but we lacked a "feedback coach" to observe our interactions and help us adjust. We built CoachLens AI to be that supportive, expert eye—helping parents bridge the gap between learning and doing.

🎯 What it Does

CoachLens AI is a multimodal coaching tool that analyzes parent-child interactions to provide actionable feedback.

  • Seamless Input: It allows users to upload video and audio covering their interactions with their kids. This eliminates the tedious process of manually writing out descriptions or transcribing dialogue for traditional text-based AI tools.
  • Automated Expertise: With built-in, specialized parenting prompts, the system automatically generates expert feedback, saving users the hassle of learning complex prompt engineering to get the right advice.
  • Immediate Reinforcement: Validates when a parent successfully uses positive reinforcement.
  • Validation Gaps: Gently points out moments where a parent might have jumped to "fixing" a problem before "validating" the child’s feelings.
  • Actionable Insights: Provides a constructive "post-game" analysis of a parent's tone, body language, and response timing to help them improve their skills over time.

🛠️ How We Built It

The project is powered by Google Gemini via Google AI Studio, leveraging its industry-leading long-context window and native multimodality.

  • Multimodal Analysis: We utilize Gemini’s ability to "see" and "hear" simultaneously, allowing the model to interpret non-verbal cues from the baby and the parent's vocal tonality.
  • Advanced Prompting: We used structured system instructions to ground the AI in evidence-based child development frameworks.
  • Timeline Mapping: We implemented a "Timeline Sniper" feature that anchors AI feedback to specific timestamps in the video, making the advice contextually relevant.

🚧 Challenges We Ran Into

  • Data Friction: Recording and uploading high-resolution video is a manual hurdle. To solve this, we optimized the pipeline to support audio-only analysis and implemented automated compression to reduce upload latency.
  • Feedback Specificity: Initial AI responses were too general. We overcame this by refining our prompt architecture to focus on "micro-interactions," forcing the AI to cite specific seconds in the footage to justify its feedback.

🏆 Accomplishments That We're Proud Of

We are most proud of the personal impact: this tool has already started guiding our own daily interactions with our son. Seeing an AI provide the same level of nuance as a human parenting coach—but with 24/7 availability—convinced us that this could foster a more emotionally resilient next generation.

💡 What We Learned

We learned that AI’s greatest value isn't just providing information—it's filling the expertise gap in real-time. However, for AI to truly rival a live coach, the input must be "ambient." We realized that reducing the friction of data collection (moving from manual uploads to "always-on" assistance) is the final frontier for AI coaching.

🚀 What's Next for CoachLens AI

  • Ambient Integration: Integrating with smart home ecosystems (like Google Nest) for automatic footage capture and analysis without needing a smartphone.
  • Live AR Coaching: Developing a "Live Mode" where parents can use a phone holder or wearable tech to receive real-time haptic or audio cues (e.g., a gentle nudge to "Remember to validate feelings first").
  • Emotional Trend Tracking: A dashboard to visualize a parent's growth in "patience scores" and "validation frequency" over months of use.

Built With

Share this project:

Updates