Korva is a scroll of the internet’s best ideas.
The AI pod
The change came from a pod of 30 of Uber's best AI engineers paired with business people and people in the finance and administration functions. They go process by process and start from the ground up, asking how AI would improve each one. Cutting the pricing-allocation process from 15 hours to two gives someone about two days of their time back.
Finance and marketing
The finance team constantly re-forecasts every inch of the business, and that process went from eight hours of work to two. The extra time adds a layer of nuance to the forecasts, so they are more precise. Marketing QA went from two weeks to two days.
Capturing the savings
The savings are hard to pin to a headcount number, because the freed hours get filled with other, presumably high-value work. The way to capture the efficiency as cost is in target setting. If AI makes employees 10, 20, or 30 percent more efficient, next year's headcount can stay flat, rise 2% instead of 10%, or fall 5%.
Why does a forecast take hours to rebuild?
A re-forecast starts by pulling actual results from ledgers, bookings systems, and cost reports, then reconciling them into one set of numbers. Those steps are repetitive and rule-based, which suits software that can read tables and write queries. Analysts then adjust assumptions for each business line and region. Finance teams commonly run these cycles monthly or quarterly.
Why do productivity gains take time to show up?
Economist Robert Solow wrote in 1987 that the computer age was visible everywhere except in the productivity statistics. Measured productivity growth in the United States picked up in the late 1990s. Economist Paul David argued in a 1990 paper that electric motors took decades to raise factory output because factories had to be redesigned around them.
More from The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
