Korva is a scroll of the internet’s best ideas.

The overhang

A model that can do remarkable things sits mostly unused if a person's instinct is to ask about the weather in San Francisco. Closing that gap is largely a product problem: finding on-ramps that guide people to the features that fit them.

Keeping up

Keeping up is hard even inside the company. Opus 4.5 unlocked a whole batch of capabilities beyond Opus 4's, which meant testing what the new model could do, learning which on-ramps worked, and shipping a new flow. By then the next model may already have unlocked newer capabilities, and those learnings go stale.

What still holds

One growth lesson still holds: some of the highest leverage comes from matching the right feature to the right user. The first day or two of the product experience is among the biggest levers on long-term retention, and its importance has grown exponentially.

Why do people use new tools like old ones?

People map a new technology onto the one it replaces. Early films were often recordings of stage plays, and early websites copied the layout of print brochures. Media theorist Marshall McLuhan wrote in 1967 that we look at the present through a rear-view mirror and march backward into the future. A chat window that resembles a search box invites search-box questions.

Why do prompt tricks go stale after releases?

Workarounds built for one model's weaknesses lose their purpose when the next model removes the weakness. Early chatbots needed prompts that told them to reason step by step. Reasoning models released from 2024 onward, such as OpenAI's o1, do that on their own, which made manual chain-of-thought prompting less necessary.

Why does the first day shape retention?

Most people who install an app stop using it quickly, so what happens in the first session decides who remains. Data from the analytics firm Quettra, published by investor Andrew Chen in 2015, found that apps lose on average about 77 percent of their daily active users within three days of install. Early use also sets habits, since people repeat the first behaviors they find useful.

More from Lenny's Podcast: Product | Growth | Career

Business & StartupsBet bigger when value is exponentialAnthropic's growth team spends more time on big swings than small optimizations, because the value of AI products may grow 100 to 1,000 times in two years.
Enjoyed it?