Developer Chats – Pavel Shchekotov

Episode Summary

Executive Summary: Pavel Shokotov explains how agency work sharpened his product thinking and prepared him to build Sophie Connects, a voice-first consumer AI product aimed at older users. He discusses designing around phone calls instead of screens, the operational challenges of latency, voicemail detection, retention, and cost control, and argues that AI’s power lies in execution speed while humans still own creative judgment and product responsibility.

Main Topics: Agency experience vs. founding engineering (Priority: 5/5): Pavel contrasts client-service agency work with startup ownership, emphasizing how founding engineers must connect product decisions to business outcomes rather than simply implement requests. Voice as the primary interface (Priority: 5/5): The product bets on phone calls and conversational AI instead of screens or swiping, driven by user familiarity, dating-app fatigue, and a desire to differentiate the experience. Building a no-UI consumer product in production (Priority: 5/5): He explains the engineering and product difficulties of managing conversational flow, latency, call cost, SMS costs, and coordinating both parties on the line without a visual interface. Designing for older demographics (Priority: 4/5): The team targets users in their 40s, 50s, and 60+, which reduces competition but requires careful UX, larger fonts, and stronger trust-building because of lower technical fluency and scam awareness. Production edge cases and voicemail detection (Priority: 4/5): A seemingly simple problem—detecting voicemail—turned into a major systems challenge that required a custom multi-layer detection approach because off-the-shelf tooling was unreliable. AI’s limits and rapid evolution (Priority: 4/5): Pavel argues AI still lacks creative thinking and human responsibility, but people underestimate how quickly capabilities improve, requiring constant reassessment of what AI can and cannot do. Future of AI-native consumer products (Priority: 4/5): He predicts dynamic, personalized user interfaces that adapt to each user’s needs, enabled by lower software engineering costs and AI-driven product customization.

Key Arguments: Agency work teaches product thinking quickly because engineers see different companies’ priorities, trade-offs, and failure modes across many projects in a short time. Founding engineers must think holistically: product changes affect marketing, retention, and user acquisition, not just code quality. Voice is a compelling primary interface for older users because it matches existing habits and avoids the fatigue associated with swipe-based dating apps. Removing the screen shifts the challenge from UI design to conversation management, cost optimization, and orchestration of outcomes within each interaction. Phone-call and SMS based products have real marginal costs, so every extra word, token, or minute affects business economics. Older users are underserved but require more usability attention, clearer visuals, and stronger trust signals. Voicemail detection was more difficult than expected and had to be rebuilt from scratch because vendor flags were too inaccurate. AI should be seen as a fast-moving executor, not a substitute for creative product judgment or accountability. Future consumer apps may become dynamically personalized, with different users seeing different versions of the same core product.

Data Points: Target age range: around 60 years old - Initial age barrier for the product’s user base Average user age in early testing: around 65 - Early exploration and product testing users Call cost: about $1 per call - Used to illustrate the cost sensitivity of prolonged conversations SMS billing unit: charged by segments - Explains why shorter text messages can materially reduce cost Timeframe referenced for AI progress: half a year ago vs. today - Used to show how quickly AI-generated content quality is improving Model examples: Clink 3.0, Sedence 2.0 - Examples cited to illustrate rapid advances in AI-generated content tools

Pivotal Quotes: "your success equals product success" — Pavel Shokotov: Describing the mindset shift from agency work to founding engineer responsibility "AI lacks creative thinking, which is a crucial thing when you build a completely new product" — Pavel Shokotov: Explaining where AI is still overestimated "we had to rebuild it and create from scratch" — Pavel Shokotov: Referring to voicemail detection after off-the-shelf handling proved unreliable

Implications: Builders should treat AI as a fast-evolving tool, not a finished solution, and design products around real user habits and economics. Voice-first interfaces may open overlooked markets, especially older users, if teams can solve trust, latency, and orchestration.