In this interview, Joel shares meaningful applications of AI in his line of work and distinguishes what’s real and what’s hype in the world of AI.

kea: Thank you for taking the time out of your schedule to discuss the business—or busyness—of AI!

Joel: Of course!

kea: Let’s start with your day-to-day activities. What projects are you working on, and what excites you about them?

Joel: Every day is different at a startup, and this is especially true in the world of AI, which evolves daily. With tens of thousands of tech companies building AI, kea is on a high-speed racetrack to optimize and advance the application of AI. That alone is exciting, but what excites me most at kea is leveraging Voice AI to make the lives of restaurant workers easier. Recently, I helped develop a product tailored for smaller brands, not just giants like Pizza Hut, which feels tangible and impactful. During my time here, we’ve also refined kea’s voice to be human-centric and realistic, keeping pace with the latest AI advancements and making it easier for restaurants to adopt our technology.

kea: How do you distinguish between what’s real and what’s hype in the AI world?

Joel: As a society, we need to recalibrate our expectations to match reality. Arthur C. Clarke once said, “Any sufficiently advanced technology is indistinguishable from magic.” In some ways, we experience that magic with Generative AI and human-like voices that are indistinguishable in tone and articulation. However, slapping the AI label on anything with a minor resemblance to AI—like a Help Center chatbot that’s a search engine in disguise—is very limited in the grand spectrum of what AI can achieve. Ultimately, it’s important to distinguish the meaningful and effective applications of AI from the superficial ones. 

kea: What ethical considerations do you take into account when developing Voice AI technologies? 

Joel: Our goal is to make Voice AI as human as possible so that restaurant guests have a seamless experience. That being said, there’s an argument to be made about the ethics of creating voices that are indistinguishable from humans. If it is indistinguishable, it’s the company’s responsibility to be transparent so users are informed. It’s also crucial to bring awareness and consent to the world of Voice AI and avoid exploiting other people’s voices or creativity at all costs. 

kea: Let’s talk about autonomous ordering. What makes it possible, and what’s its potential right now?

Joel: Creating AI with logical understanding is key for autonomous ordering, especially given the complexities of restaurant menus. For example, ensuring that a salad doesn’t come with pizza toppings is a unique challenge for AI. Using a dynamic medium like Voice AI is difficult, but kea excels at it. We leverage vast amounts of data to optimize our AI’s performance and structure conversations so our system can handle them without error. The ability to transform free-flowing, dynamic conversations into structured systems like APIs is what sets kea’s autonomous ordering apart. 

kea: What advice would you give to someone aspiring to work in Voice AI or project management in tech?

Joel: Get hands-on experience and practice prompt engineering. Generative AI changed linguistics in a similar way that calculators changed math; you don’t have to do the math manually, but you still need to understand the fundamentals. The most important skill in this role is being able to articulate your ideas accurately with prompts. Secondly, stay updated with the latest news and research regularly to understand the evolving landscape. Striking a balance between making progress and not frustrating engineering teams with constant pivots is crucial.

kea: If you could develop a dream Voice AI feature, what would it be and why?

Joel: I’d like to see Voice AI get better at turning unstructured conversations into structured outputs. It should be able to anticipate the calibrations you forgot to input and ask if those are important for it to know. AI should check with the human designer to ensure it can produce more accurate and useful information, aligning with what the user originally intended.

kea: Amazing! Thank you so much for sharing everything and best of luck with the next project!