Ted Johnson is co-founder of JoinIn.ai, where he leads technical direction and system architecture for AI designed to participate in human conversations. His defining argument: Modern AI still forces people to submit carefully packaged requests when intelligent systems could instead follow the interruptions, ambiguity, and shifting context of ordinary discussion.
Johnson studied computer science at the University of Minnesota, focusing on human-computer interaction and collaborative systems, and holds a master’s degree in computer science. He spent approximately 25 years building enterprise software, collaboration systems, and financial-services technology. He served as vice president of architecture at Ameriprise Financial’s Columbia Threadneedle Investments before becoming chief architect at Transamerica, working across insurance, retirement, distribution, cloud migration, and enterprise data platforms.
In 2026, Johnson announced JoinIn.ai with co-founder Brendan Foote, bringing his enterprise-architecture background to multiperson conversational systems.
- The prompt is still a punch card. Johnson argues that prompting preserves batch computing: Assemble a request, submit it, wait, inspect the answer, and repeat. His critique of prompt-centered interfaces treats prompt engineering as compensation for an outdated interaction protocol; transcribing speech into alternating turns does not resolve the underlying constraint.
- Channel, expression, and protocol. He distinguishes how information travels, how much meaning an interface accepts, and when participants can contribute. Language models vastly expanded expressive range, but the submit-and-wait protocol remained. Genuinely conversational systems need responsiveness, interruption handling, and shared context—not merely microphones.
- Conversation-native AI. Johnson’s approach incorporates speaker-reference tracking, uncertainty, collective decisions, and whether anyone is addressing the assistant. A useful system should recognize immediate versus future requirements, seek confirmation before consequential changes, and remain silent when intervention would be unhelpful.
- Pulse and Baton. JoinIn.ai’s Pulse meeting assistant and Baton turn-taking engine apply these principles through speaker-aware transcription, agenda tracking, contextual answers, contradiction alerts, action-item extraction, and decisions about when an assistant should speak or wait.
