Project Overview:
We are building "The Digital Closer," a high-performance AI co-pilot for intake and sales teams. The goal is to provide real-time, word-for-word scripts to sales reps on live Zoom/Google Meet calls the moment a prospect raises an objection (Price, Timing, Spouse, DIY, etc.).
We have the "Closing DNA" (proven legal sales scripts). We need the engineer to build the real-time delivery system.
Key Responsibilities:
Meeting Integration: Implement Recall.ai to deploy a recording bot into live video meetings.
Ultra-Low Latency Transcription: Integrate Deepgram’s Live Streaming API to achieve sub-second speech-to-text.
Semantic Trigger Engine: Build the logic (via GPT-4o or Claude 3.5) to detect specific objection types in real-time.
Chrome Extension Overlay: Develop a Manifest V3 Chrome Extension that "whispers" the correct script to the rep via a transparent floating UI.
Security: Ensure the pipeline is SOC2-ready (data must be encrypted and ephemeral).
Required Tech Stack:
APIs: Recall.ai, Deepgram (Streaming), OpenAI/Anthropic.
Frontend: React/TypeScript, Chrome Extension Manifest V3.
Backend: Node.js or Python.
Data: Vector databases (Pinecone/Weaviate) for script retrieval (RAG).
The Ideal Candidate:
You have built a "Live-Assist" or "Conversation Intelligence" tool before.
You understand that 1 second of lag is too long in a sales call.
You can show me a portfolio item involving a Chrome Extension or a Meeting Bot.
How to Apply:
Start your proposal with the word "CLOSER" to prove you read the brief.
Briefly describe your experience with real-time streaming APIs (specifically Deepgram or Recall.ai).
Estimate the timeline for a "Minimum Viable Product" (MVP) that handles just one objection trigger.