The Manual-to-Scale Automation Ladder
by Brian Tolkin • Head of Product and Design at Opendoor
Brian is the Head of Product and Design at Opendoor. Previously, he was Employee #100 at Uber, where he led the global launch of uberPOOL and established the original Product Operations function.
🎙️ Episode Context
Brian Tolkin shares deep insights on building products that bridge the physical and digital worlds, drawing from his experiences scaling Uber and Opendoor. He discusses the evolution of operational processes into scalable software, how to run non-threatening product reviews, applying Jobs-to-be-Done for low-frequency use cases, and how to rigorously experiment when you lack high transaction volume.
Problem It Solves
Determining when to solve a problem with human operations versus when to invest in engineering resources to automate it.
Framework Overview
A staged approach to scaling complex physical-world processes. Instead of building software immediately, the company validates the process manually, scales it via batching, and only builds technology when human scaling breaks.
🧠 Framework Structure
Stage 1: High-Touch Manual (Do things...
Stage 2: Process Batching. Increase e...
Stage 3: Content Scaling. Use static ...
Stage 4: Tech-Enabled Verification. B...
Stage 5: Reallocate Ops. Once automat...
When to Use
When launching new verticals, entering new markets, or building features that require heavy operational fulfillment.
Common Mistakes
Building expensive technology automation before the operational process has been proven manually or before the volume necessitates it.
Real World Example
Uber's driver onboarding evolution: It started with 1:1 in-person meetings, moved to small classrooms, then video sessions, and finally automated OCR technology for document verification once volume exploded.
Computers are deterministic, but humans aren't... building products that have a little bit more flex or a little bit more fail safes in case those things happen becomes a little bit more of a paramount.
— Brian Tolkin