Front-End Fire 141: When AI Belongs in Products with Dan Wilson & Kyler Krenek
Apr 21, 2026
Kyler Krenek, cofounder of Vowel, explains Vowel’s voice-AI platform and how to integrate low-latency voice into web apps. Dan Wilson, Principal Partner at DataCurl, explores where AI belongs in products and design choices for adoption. They discuss voice vs chat UX, client-side RAG, safety controls like confirmations and approvals, and measuring AI value and ROI.
AI Snips
Chapters
Transcript
Episode notes
Voice Lets Users Speak Their Stream Of Thought
- Voice lets users stream their thoughts directly to the system, reducing friction from editing typed prompts.
- Kyler observed dictation tools and a foot-pedal workflow that shifted him from typing to thinking aloud, enabling richer prompts.
Model Costs Matter Don’t Ignore Token Economics
- Do calculate AI costs and ROI before integrating features; token and runtime costs change architecture decisions.
- Dan spends time with product owners to quantify budgets because AI's variable cost model can surprise stakeholders.
Choose Voice Where Convenience And Safety Matter
- Do pick voice for contexts where safety, convenience, or reduced distraction matter, like driving or field sales.
- Dan gave a salesperson example: let them speak orders in their car instead of typing later to speed revenue activities.