UNASKED

main dasboard

Inspiration

People often ask questions, but what they really want to know is hidden behind fear, anxiety, or uncertainty. For example:

“Should I take a drop year?” is often really: “Am I afraid of failing again?”

This inspired us to build UNASKED, an AI system that doesn’t just answer questions — it uncovers the unasked question behind them.

We wanted to combine psychology, natural language understanding, and actionable guidance into one tool that helps people gain clarity in life decisions.

What it does

UNASKED is an AI system that analyzes not just what users ask, but what they really mean and feel behind the question. Instead of giving generic answers, it uncovers hidden thoughts, emotions, and intentions, then guides users toward clarity and action. 🔍 Core Capabilities 🧠 Hidden Question Detection Identifies the real unspoken question behind the surface question. Example: Surface: Should I take a drop year? Hidden: Am I afraid of failing again? 😶 Emotion Detection Detects underlying emotions like: Anxiety Fear Confusion Hope Frustration Returns confidence scores for emotional analysis. 🎯 Intent Classification Classifies the user’s intent, such as: Career confusion Academic stress Self-doubt Decision-making Validation seeking 🔄 Reframed Perspective Rewrites the question to help users think differently. Helps users see problems from a clearer, less biased viewpoint.

How we built it

UNASKED is built as a full-stack AI system: 🔹 Backend (Python + Flask) Text processing and AI prompt engineering Emotion detection and intent classification Hidden-question inference Structured JSON output with: Hidden Question Emotion + Confidence Intent Reframed Perspective Actionable Next Steps Reflection Questions Clarity & Emotional Intensity Scores ELI5 Summary

Challenges we ran into

1️⃣ Extracting Hidden Intent User questions are ambiguous. Designing prompts to infer unasked questions without hallucination was difficult. 2️⃣ Emotion Detection Mapping natural language to emotions with confidence scores required iterative tuning and validation. 3️⃣ Structured AI Output Ensuring the AI always returns clean JSON fields (for frontend UI) was challenging. 4️⃣ UX Design Presenting deep psychological insights in a friendly and non-overwhelming way required multiple UI iterations. 5️⃣ Responsible AI We added mental-health disclaimers and reflection prompts to avoid harmful advice.

Accomplishments that we're proud of

What we learned

Building UNASKED taught us: How user questions contain latent intent and emotional signals Practical NLP techniques for emotion detection and intent classification Designing AI outputs that go beyond text and provide structured insights Frontend UX for conversational AI interfaces How to integrate LLM APIs into a full-stack system The importance of responsible AI and mental-health awareness We also learned that clarity often comes from asking the right question, not just giving the right answer.