🚀 Consent Lens

🧠 Tagline

"Read the room before you hit send."
AI-powered tone checker for respectful, boundary-aware communication.

🧩 Inspiration

Every day, people send messages with good intentions — but unclear tone, emotional pressure, or guilt-tripping can lead to hurt, misunderstanding, or regret.
There are tools to fix spelling, grammar, and even tone. But nothing helps you check:

“Is this message emotionally respectful? Could it pressure someone unintentionally?”

Consent Lens was born from that gap.
It’s designed to help people communicate more clearly and more kindly — especially in moments that really matter: breakups, confessions, follow-ups, and boundary-setting.

🛠️ What It Does

Consent Lens helps users reflect on the tone of messages — particularly around consent, emotional safety, and interpersonal boundaries.

Paste a message → click Analyze Message → get:

🎭 Tone Classification: Respectful, Neutral, or Concerning
📊 Score Breakdown:
- Clarity (0–10)
- Respect (0–10)
- Pressure Level (0–10)
💬 AI Feedback: Explains potential concerns, and how to soften or clarify the message

It acts as a pre-send reflection tool — like Grammarly, but for emotional responsibility.

❓ Why Not Just Use ChatGPT?

Yes, someone could ask ChatGPT or another AI: "Is this message respectful?" — but that’s missing the point.

Here’s why Consent Lens needs to exist:

🧠 Purpose-Built: ChatGPT is general. Consent Lens is specifically designed for analyzing tone in emotionally charged conversations involving consent, guilt, and pressure.
🧼 Frictionless: No prompt writing or context needed — just paste and click. It’s fast, private, and focused.
📊 Structured Feedback: It gives consistent scoring for Clarity, Respect, and Pressure. Chatbots give different answers every time.
🧑‍🤝‍🧑 Community-Ready: This is something a therapist, friend, or teacher could recommend. It's a tool, not a chat.

It’s not just about being right — it’s about helping people reflect before they hit send. That’s something a focused tool can offer better than a general AI.

🔧 How I Built It

Platform: Bolt.new — natural language-powered app builder
Prompt Engineering: Carefully designed prompts to analyze tone, detect coercion or guilt-tripping, and balance pressure/respect/clarity scoring
Scoring Logic: Custom deductions for emotionally manipulative phrases (e.g. "you owe me", "are you ignoring me"), visual cues (ALL CAPS, “???”), and positive language bonuses
UI Design: Created with calming pastel colors, centered layout, and a judgment-free interface

💡 What I Learned

How to build emotionally aware AI tools with low-code platforms
That subtle pressure and tone are harder to detect than obvious language
Why emotional design matters just as much as functionality
How to balance AI precision with human nuance

🧱 Challenges I Faced

Misclassification: The model marked some guilt-tripping phrases as “Respectful.” I improved phrase detection logic and tuned pressure scoring.
False positives from neutral language: I added balance by reintroducing score bonuses for clearly respectful language like “if that’s okay” or “no pressure.”
Design tone: Making something honest but not judgmental required iterating feedback phrasing and visual design.

🚀 What’s Next

Add rephrase suggestions: “Here’s a kinder version of your message”
Let users compare two versions: “Which one sounds more respectful?”
Turn it into a browser extension or email plugin
Train on real-world anonymized data (with consent) to improve subtle pressure detection