Accelerating product development in the AI age: How Faire developed product builders
When AI tripled Faire's engineering output, product/design became the bottleneck—so instead of hiring more PMs, they trained engineers to own features end-to-end, eliminating handoffs entirely.
Read Original Summary used for search
TLDR
• AI 3x'd engineer throughput (complexity-weighted PRs), making traditional PM/design processes the new bottleneck—not coding speed
• Solution: "Product builders"—engineers who can take a customer problem, design the solution, and ship it end-to-end without handoffs
• Built a "Complexity Tier Framework" to match feature complexity to engineer skill level, plus bi-monthly assessments and PM/designer coaching
• After 2 months: 71% of engineers shipped end-to-end features, 65% rated Capable/Strong in product thinking (up from 41%)
• Real impact: Engineers now spot opportunities themselves (like fixing recommendation personalization for 1.5% order lift) instead of waiting for specs
In Detail
Faire's engineering team hit an unexpected problem when AI coding tools tripled their output: the traditional product development process—with its handoffs between PM, designer, and engineer—became the bottleneck. Their counterintuitive solution wasn't to hire more PMs and designers to clear the backlog, but to develop engineers into "product builders" who could own features end-to-end. A product builder is an engineer with enough product and design thinking to take a scoped customer problem, design a solution, implement it, and ship it—no handoffs required.
The implementation required real organizational change, not just new tools. They created a "Complexity Tier Framework" that matches feature complexity to engineer capability (Tier 1: familiar UX within one team, Tier 2: cross-team coordination, Tier 3: novel customer problems). They built coaching infrastructure with skill toolkits, live training, and enlisted senior PMs/designers as mentors. Engineering managers assess each product engineer's product and design thinking capabilities every two months using concrete definitions (Developing/Capable/Strong). They were explicit about the stakes: AI is changing how software gets built, and everyone needs to adapt or risk becoming irrelevant.
The results came faster than expected. Two months in, 71% of product engineers had shipped a feature end-to-end, and the percentage rated Capable or Strong jumped from 41% to 65%. More importantly, engineers started spotting opportunities themselves—like one who noticed beta recommendations weren't personalized enough, wrote the PRD, aligned with their PM, and shipped improvements that lifted order volume 1.5%. The real insight: when AI removes one constraint (coding speed), the bottleneck shifts to handoffs and coordination. The solution isn't more specialists—it's eliminating the handoffs by growing engineers' capabilities.