Korva is a scroll of the internet’s best ideas.
Startups win by caring more
One concentrated bet
Microsoft and Google have trillions of dollars more in resources, 100,000 more people, and infinitely more brand name. A startup that does everything loses to them. A lasting business comes from narrowing to one specific thing, making a concentrated bet that turns out to be right, and executing tightly.
Caring more than rivals
Daniel Ek gave the Spotify version when people asked why it could stand against Apple Music and YouTube: Spotify would simply care far more about music than they would. Cognition applies the same reasoning to software. The work is building software end to end at a place like Goldman Sachs or Mercedes-Benz, which means understanding an existing codebase, collaborating with the humans on it, plugging into a ticketing system, and letting the agent test its own code locally.
The big-company objection
The objection in 2023 and 2024 was that Microsoft already had GitHub Copilot, VS Code, and the OpenAI partnership. The same argument was made years earlier about Datadog, Snowflake, and Databricks, since the clouds care about observability too. Millions of problems exist, the world keeps changing, and there are many niches to own.
Why do giants lose to focused startups?
Clayton Christensen described this pattern in his 1997 book The Innovator's Dilemma. Large firms listen to their biggest existing customers and put resources into improving current products. A small entrant can focus on a narrow need that the big firm treats as too small to bother with. As that niche grows, the entrant can move into the larger market.
How does an agent plug into a ticketing system?
Ticketing tools such as Jira and Linear expose their data through application programming interfaces. An agent can read a ticket's description, post comments, and update its status the same way a teammate would. The ticket becomes the unit of work the agent is handed. Engineering teams already organize their work around tickets, so the agent fits into a process that exists.
Why do cloud providers not own observability?
Amazon, Microsoft, and Google each sell monitoring tools bundled with their own clouds. Customers who run on several clouds want one view across all of them. Datadog built its business on that neutral position, and it went public in 2019. A neutral vendor can sell to companies that would hesitate to rely on a cloud rival's tool.
More from David Senra
