Conversational AI prototype
AI Voice Companion
Conversational AI that talks like your smartest friend, and remembers the last call.
- Builder
- Karnveer Singh
- Status
- Prototyping
- Repository
- Private
- Demo
- Available on request
Purpose
The AI Voice Companion is a conversational AI prototype by Karnveer Singh built for mentorship, brainstorming and companionship rather than voice commands.
It is designed around continuity: the assistant should remember previous conversations and pick up threads instead of starting cold.
Architecture
A realtime speech pipeline handles low-latency listening and speaking so conversation stays natural.
A persistent memory store keeps facts and open threads between calls.
A reasoning layer decides when to be proactive — surfacing a reminder or a follow-up question rather than waiting to be asked.
Challenges
Latency: natural conversation collapses above roughly a second of delay, which constrains every model choice.
Interruption handling and turn-taking that feels human.
Deciding what is worth remembering from spoken conversation, where most content is filler.
Technologies
- Realtime voice
- Speech-to-text and text-to-speech
- Memory systems
- Python
Research
Feeds directly into research on AI memory and human-AI collaboration.
Related research notes
- AI Memory: why context windows are not memory
Notes on building durable AI memory — what is worth storing, how retrieval decays, and why bigger context windows do not make an assistant remember.
- Human-AI collaboration: one voice across many models
Why a portable reasoning identity beats model loyalty, and how a boot prompt keeps behaviour consistent across AI systems.
Frequently asked questions
- What is the AI Voice Companion?
- It is a conversational AI prototype built by Karnveer Singh that combines realtime voice, persistent memory and proactive reasoning for mentorship and brainstorming.
- Can I try the AI Voice Companion?
- It is a private prototype. A demonstration is available on request.