Blog
Thoughts and insights on building conversation-native AI
Why a Group-Meeting Company Runs a One-on-One Benchmark
End-of-turn detection is the floor under everything Baton does. Why we measure it on someone else’s two-party benchmark, where LiveKit’s board puts Baton — second of thirteen — and what that number still cannot tell you about a room with four people in it.
Read article →
Why We Built Baton
Frontier voice models give good answers with terrible timing. Baton is the turn-taking engine that decides when your AI speaks in a group — and when it stays silent. Why we built it, and why it is named after a relay hand-off.
Read article →
The Prompt Is Still a Punch Card
The intelligence felt radically new. The interface felt strangely old. AI should not just make software smarter — it should change the interface between humans and computers.
Read article →
Conversation Is Thinking Output in Real Time
Humans do not simply communicate ideas through conversation. We discover them there. Conversation is thinking output in real time.
Read article →Founder Essay: Conversation is the Interface
Why we're building JoinIn.ai: The intelligence and power of AI is still stuck in prompt and response. Real human life happens in conversation—and that's the next interface AI must master.
Read article →
Our Vision for AI as a Meeting Participant
Imagine a world where you didn't have to think about how you had to interact with AI. Learn how JoinIn.ai is building conversation-native AI that participates in meetings like a teammate.
Read article →
AI Conversation Training Data
Exploring the importance of high-quality conversation data for training AI systems that can truly understand and participate in multi-speaker interactions.
Read article →