Voice AI: What Works, What Breaks and What Comes Next

Darius Smith builds AI-powered systems that connect cloud infrastructure, software engineering and modern AI workflows. His experience includes LLMs, RAG, vector databases, agents, automation, Python, FastAPI, React and end-to-end AI products across sectors including healthcare, construction, logistics and SaaS. Darius also built the current ViVo Pulse app for WattNext and created the voice agent that can be experienced on our website.

Taylor Blake leads AI Labs at Degreed, developing voice-AI capabilities for enterprise learning, coaching and roleplay. His work includes ElevenLabs and OpenAI technologies, with experience navigating the practical governance and approval requirements of large organisations.

Jillyn Johnson is a founder, AI voice designer and specialist in the last-mile testing of conversational voice systems. Through Build Solutions, she helps organisations move from an impressive voice AI demo to something people can use and trust in real customer and operational environments. Her work spans voice-agent design, conversation analysis, AI readiness, multilingual systems, onboarding, lead qualification and voice-based tools for elder companionship and language translation. What sets Jillyn apart is her hands-on testing of both the agent and the user experience: she listens for unclear questions, weak context, awkward handoffs, language and accessibility issues, as well as problems with access, privacy and security. She also tests agents autonomously and analyses transcripts to identify the critical issues that automated checks can miss. With more than 20 years of wider business and marketing experience, Jillyn brings a practical, commercially grounded perspective on what works, what breaks and what must be fixed before voice AI is ready to rely on.

Andrew Ilingin is a serial entrepreneur and founder of Dialog AI, an AI transformation company building and operating voice agents for K-12 education and other high-volume customer workflows. Dialog’s agents are in production across four continents, handling use cases such as lead qualification, enrolment, reactivation and rescheduling across multiple languages. Andrew brings a grounded view of what happens when Voice AI moves beyond the demo: how to choose a job the agent can genuinely do, connect it to existing systems, measure commercial outcomes and maintain quality as usage scales. Previously, he founded six EdTech startups and led an eight-figure acquisition of Yaklass, a K-12 platform used by 11 million users and 40,000 schools. Across eight startups and two exits, his focus has been scaling early-stage ventures, building high-impact teams and turning emerging technology into operating businesses. This gives him a direct perspective on what works, what breaks and what comes next in Voice AI.

Marco Bettiolo is co-founder and CTO of Sensay, where he builds voice-first AI systems that capture, transfer and retain critical organisational knowledge. Sensay’s AI interviewer, Sophia, conducts natural offboarding conversations, drawing out the tacit knowledge people carry and making it available to teams as searchable, reusable digital knowledge. Marco brings more than 20 years of experience in software design and development across education, events and learning technology, together with a practical understanding of what it takes to make AI work inside real organisations. His perspective on Voice AI is grounded in the gap between a convincing conversation and a dependable product: designing agents that can handle context, ask useful follow-up questions, preserve nuance and earn trust around sensitive information. At this panel, he will explore what Voice AI can genuinely do today, where deployments break down and how builders should think about reliability, handoff and the next wave of capability.