geekyroshan
Back to Projects
AllysAI Consulting Agent
Conversational AI

AllysAI Consulting Agent

Voice + text AI consultant with sub-4-second latency, barge-in support, and real-time lead qualification.

About the Project

A voice-based AI consulting agent that helps potential clients assess their AI readiness through natural conversation. The key engineering challenge was latency -- the pipeline streams LLM output, sends the first sentence to TTS immediately, and synthesizes the rest concurrently while the first sentence plays. This gets audio to the user in under 4 seconds. Supports barge-in: if the user starts talking mid-response, the agent stops and listens.

Key Features

  • Sub-4-second end-to-end voice pipeline: STT + RAG + LLM + TTS
  • First sentence plays immediately while rest synthesizes concurrently
  • Barge-in support -- interrupt the AI mid-response and it listens
  • Pinecone RAG for knowledge-grounded responses
  • Real-time metrics: readiness score, fit score, ROI projections
  • Automatic lead capture and qualification from conversations

Impact

Production voice agent with sub-4-second latency serving AllysAI's sales pipeline. Demonstrates real-time streaming, concurrent audio synthesis, and conversational lead qualification.

Tech Stack

TypeScriptReactExpressWebSocketWhisperGPT-4o-miniElevenLabsPineconeSQLite

Metrics

<4s end-to-end voice latency
Barge-in support
Live readiness scoring
RAG-powered

Interested in this project?

Let's discuss how I can build something similar for you.