Kite: 500,000 monthly developers who would not pay

What it is

Kite was a startup that used AI to help developers write code. In its farewell post, founder Adam Smith says the company ran from 2014 to 2021 and that it no longer supports the Kite software (Kite farewell post, archived). The live kite.com domain no longer hosts the product, so we cite an archived copy of the post.

Timeline

• 2014: Kite starts as a company using AI to help developers write code (Kite farewell post, archived).
• 2019: Smith writes that Kite reached product-market fit, five years after starting, after "many iterations and heavy engineering lifts" (Kite farewell post, archived).
• After 2019: Kite grows its user base to 500,000 monthly-active developers "with almost zero marketing spend" (Kite farewell post, archived).
• 2021: Kite's active work ends. The team explored a pivot to code search, but Smith says the team was too tired after seven years to pursue it (Kite farewell post, archived).
• 16 November 2022: Smith publishes the farewell post and points readers to the code on GitHub (Kite farewell post, archived).
• Afterwards: the public repository describes itself as a version of the main Kite repo, adapted for publication by replacing private information (kiteco-public on GitHub).

Why it worked

Here "worked" means adoption, not a business. Kite got real usage.

• It built a hard product with a strong team first. Smith says Kite started by building "a world-class engineering team" because the product was very difficult to build (Kite farewell post, archived).
• Free distribution spread among developers. The 500,000 monthly-active developer figure came "with almost zero marketing spend", according to the founder (Kite farewell post, archived).
• The technical work outlived the company. The public repo includes a GitHub crawler, Python package exploration and type-induction pipelines (kiteco-public on GitHub).

What to learn

1. Test willingness to pay before you scale free usage. Kite's order was team, product, distribution and then monetization, and Smith says it "took too long to figure that out" (Kite farewell post, archived).
2. Know who holds the budget. Smith's diagnosis is that individual developers do not pay for tools, and managers wanted discrete new capabilities rather than a speed-up (Kite farewell post, archived).
3. Being early is a risk, not a moat. Compare how later AI coding tools were funded and sold in Windsurf and Devin.

Risks and what didn't work

• The product did not monetize. In Smith's words: "Our 500k developers would not pay to use it" (Kite farewell post, archived).
• The technology was early. Smith says Kite was "10+ years too early to market" and fell short of the 10× improvement he thought it needed (Kite farewell post, archived).
• Founder fatigue. The pivot to code search was not pursued because the team was tired after seven years of startup stress (Kite farewell post, archived).
• The user number is company-reported. The post gives no revenue, funding total or cost figures, and we do not estimate them.