Overview

Former Google Labs lead Clay Bavor shared why he left his “dream job” to co-found Sierra: the LLM inflection, a longtime co-founder, and a chance to build from first principles. He contrasted Google’s tech-first mindset with what he had to unlearn—startups must be customer-obsessed—while keeping the habit of inventing at the edge. Clay defined an agent as software you give a goal and guardrails, plus knowledge, tools, and the ability to reason and act; Sierra builds branded, customer-facing agents for sales, support, churn saves, and more (retail, healthcare, telco, consumer tech). He argued generic chatbots miss the “iceberg” of hard problems (release/process, evals, user sims, latency, voice quirks like “check my balance” → “chicken salad”), and described Sierra’s deep work on voice (a “voice sommelier,” custom voices, 30+ languages) and reliability (speculative calls, diarization). Sierra prices on outcomes—only charging when the agent actually resolves or sells—so ROI is clear and incentives align; key metrics are CSAT/NPS, resolution rate, and revenue lift. Looking ahead, he’s excited about agents as force multipliers that unlock a wave of micro-entrepreneurship across domains.