RJ Burnham has spent more than 25 years building at the edge of voice, communications, computer vision, and AI. As co-founder and CTO of Voxeo, he helped pioneer the CPaaS market and make telecom application development accessible to a much broader developer community. Today, he is applying those lessons to healthcare as the founder of Consig AI, using voice AI to improve patient outreach and communication.
In this episode, RJ joins Seth Narayanan to explore how voice technology evolved from rigid directed-dialogue systems to large language models, why the newest technology is not always the right technology, and what enterprise leaders should understand about moving AI from impressive demos into practical production systems.
“There is a chance to actually make an impact in people’s lives.”
RJ’s career has moved through several major technology waves, from telecom and CPaaS to computer vision, neural networks, and today’s generative AI. After helping build Voxeo into a communications platform used by a large developer ecosystem, he eventually returned to voice with a new purpose: improving healthcare communication. At Consig AI, that means helping care teams reach more patients without simply replacing human conversations.
“The technology is the second part.”
One of RJ’s most enduring startup lessons is that engineers can become too attached to the technology they have built. Customers, however, are buying solutions. Long-term architectural bets matter, but founders must first understand the real problem, the user, and the outcome they are trying to create.
“Approaching it as a learner as opposed to an expert often can make you an expert very quickly.”
RJ’s unconventional education shaped a leadership style built around curiosity. His advice to founders is simple: acknowledge what you do not know, ask questions, and learn from people who understand the industry better than you do. That openness can also build more trust with customers and teams than pretending to have every answer.
“Sometimes it really is having the wisdom to go back.”
Healthcare requires natural conversations, but it also demands predictable, compliant behavior. RJ argues that the answer is often not LLMs alone, but a combination of generative AI, deterministic logic, directed dialogue, and traditional software engineering.
“It’s not about automating every conversation.”
RJ believes AI may be overhyped in the short term while still being underappreciated in the long term. Many early implementations will fail, but enterprise adoption has crossed a threshold. The priority now is finding the right use cases, delivering quality first, and then optimizing cost and efficiency.