Lucid
Full-Cycle Product Development
Read the
Room Before
It Speaks
challenge
Video calls strip away most of the nonverbal signals people rely on in face-to-face meetings. Sales reps, managers, and consultants finish calls with no idea which topics landed and which fell flat — only a vague gut feeling and a wall of meeting notes.
Approach
We layered a real-time perception engine on top of the call interface. The AI reads facial micro-expressions and vocal cadence to surface contextual nudges mid-conversation — things like "Andrew didn’t like the freemium idea" or "interest peaked at enterprise pricing." A living agenda on the right adapts as topics shift, checking off covered points and suggesting new ones based on what the AI picks up.
The Work
Lucid Call Assistant
The interface stays out of the way until it has something worth saying. A floating nudge bar delivers sentiment-based insights without breaking eye contact. The conversation plan panel tracks progress through agenda items, marks completed topics, and lets the AI inject new ones when it detects an opening. A speed-talk indicator at the bottom gives real-time pacing feedback — green when natural, red when rushing.
Nudge Insights
Real-time sentiment analysis surfaces contextual nudges mid-conversation. The AI reads facial micro-expressions to detect interest, hesitation, or disagreement — and delivers actionable cues like "interested in enterprise pricing" or "didn’t like the freemium idea."
Speech Pacing
A live pacing indicator tracks speaking speed throughout the call. Green when the rhythm is natural, red when rushing. Helps speakers stay deliberate during high-stakes conversations without breaking flow.
AI Conversation Plan
The agenda panel adapts in real time. As topics are covered, items get checked off. When the AI detects an opening or a shift in interest, it suggests new discussion points — keeping the conversation structured but flexible.
Lucid shipped a working call assistant with real-time sentiment nudges, adaptive agenda tracking, and speech pacing feedback. The product entered user testing with a group of sales teams.