💡 Inspiration
Most AI chat experiences today are reactive — they wait for users to initiate the conversation. We wanted to flip that dynamic. Our goal with Inquire.AI was to build a system that leads the interaction by offering a variety of fully audio-based experiences that feel natural, immersive, and alive. We envisioned a platform that combines the spontaneity of a podcast, the curiosity of an interactive story, and the adaptability of a human conversation — all powered by AI.
🎧 What It Does
Inquire.AI delivers audio-first, interactive experiences designed to be educational, engaging, and entertaining. Users can choose from curated scenarios where the AI doesn’t just respond — it narrates, guides, and adapts in real time.
In Educational Mode, the AI takes the role of a conversational teacher or podcast host, leading a natural back-and-forth dialogue that helps users explore new topics in an engaging way.
In Entertainment Mode, users enter generative storylines — such as an AI-led escape room — where every voice response changes the narrative path.
The result is an AI experience that feels less like a chatbot and more like an intelligent, audio-driven companion.
🏗️ How We Built It
We built Inquire.AI using modern voice and AI technologies optimized for low-latency performance and natural interaction:
LiveKit for real-time, low-latency audio streaming and session management.
ElevenLabs for ultra-realistic text-to-speech (TTS) to give the AI a dynamic, natural voice.
Deepgram for fast and accurate speech-to-text (STT) to capture and interpret user input.
Groq for lightning-fast inference, ensuring that responses feel instantaneous and conversational.
Frontend built in React, with smooth CSS animations and minimalist design for a seamless user experience.
Together, these components form a pipeline where the AI can hear, think, and respond in near real-time, maintaining a natural flow of conversation.
⚔️ Challenges We Ran Into
Real-time Synchronization: Balancing multiple asynchronous pipelines (audio input, transcription, inference, and speech synthesis) without noticeable lag was one of the toughest parts.
Natural Conversational Flow: Making the AI “lead” while still being responsive to users required careful prompt design and dynamic context management.
Voice Overlaps and Interruptions: Handling user interruptions or mid-sentence responses in a way that still felt coherent and natural was a constant design challenge.
Integration Complexity: Coordinating multiple APIs and frameworks (LiveKit, Deepgram, ElevenLabs, Groq) demanded precise control over event timing and error handling.
🏅 Accomplishments We’re Proud Of
Successfully built a real-time conversational AI experience that feels human, fluid, and emotionally engaging.
Designed a framework for AI-led dialogue, where the system guides the user rather than passively replying.
Managed to achieve low latency through optimized Groq inference and streaming audio pipelines.
🌟 What’s Next for Inquire.AI
Expanding the platform to include multi-agent storytelling, where AI characters can interact with each other and the user simultaneously.
Adding emotion-aware voice modulation, where the AI adjusts its tone and pace based on user sentiment.
Introducing customizable experiences, letting users design their own educational or narrative scenarios.
Building a mobile version for hands-free, on-the-go engagement.
Exploring integrations with spatial audio and VR environments for deeper immersion.
Log in or sign up for Devpost to join the conversation.