Meet Zac, co-founder at Flowtica, who joined the BotCast #1 to answer our usual questions on building with Voice AI.
Who are you?
I'm Zac, the co-founder of Flowtica. Our first product is a pen called Flowtica Scribe. It is a real pen, so we can write on paper and it can also record meetings. I press the button to start recording. During the meeting, if you hear something you feel is important, you can press this button called Flowmark. Once pressed, it marks that moment so AI knows, "Okay, this is the part I should care about." And if you hear something you need to follow up on, you can double-click it, so then AI uses conversations around that moment as the context and creates a to-do.
How did you end up using voice AI?
I started with the problem that I keep seeing after meetings. I think the end of the meeting is just the beginning of the work. You leave with the ideas, decisions, and follow-ups, and then you still need to move them forward. So, I started asking a simple question: "Can AI keep that context and help us with the next step?" That question brought me into voice AI.
A moment of joy and pain using voice tech.
Voice hardware is especially hard because it is a system problem. You need good audio in a real room. The system needs to know who is speaking, so we are talking about diarization. And the recording needs to transfer from your pen to your app reliably, which includes things like battery, Bluetooth, the app, and the AI backend. All of them need to work together. But also, if you do it right, it becomes part of the moat. That is the reason I value the physical product in voice AI.
What lesson would you share with other builders?
The lesson I want to share with other builders is to stay close to your user and keep iterating. Because as a founder, you need conviction, but you also need to change your mind sometimes. Do not let your ego make that hard.
Where do you see the industry in 12 months and in 5 years?
I spend a lot of time on Product Hunt. I am one of the most active hunters there. Right now, I am watching Edge AI very closely. In 12 months, I think more ASR will happen on-device. I think the phone will still be the main AI device for most of us. This can make voice products faster and it can also reduce privacy concerns. I think products will use AI both on the device and in the cloud. As for in five years, I think full duplex voice will feel more normal. I think AI will listen and speak at the same time. It also will think and use tools during the conversation, and also use visual and personal context too. At that time, I think the bigger question will be control. How do I tell AI what I want? How do I set limits? And how do I correct the AI's mistakes?
