In the Y Combinator Lightcone podcast episode What Happens When Everyone Can Build Software, the question on the table already matters to small businesses: what happens when anyone can build software on their own? The guest is Paul Copplestone, founder and CEO of Supabase.
Supabase went through Y Combinator’s Summer 2020 batch. According to YC’s episode description, the platform now has 13 million users and manages more than 50 million databases. The company is preparing for a future in which everyone can write their own software, billions of people have a personal database and AI agents generate much of the demand.
Builders are no longer just programmers
The episode’s central framing is the difference between traditional developers and a new kind of builder. A dedicated segment covers that comparison, along with Supabase’s three eras and the fourfold growth in its revenue. For small business owners the idea will sound familiar: an accountant, a hairdresser or an online retailer can now put together a working tool without hiring a software agency.
The shift does not remove the need for expertise; it moves it. Once you can build a tool yourself, the more important questions become what data it handles, who maintains it and how it stays secure. We explored the same theme in our piece on building a working environment for AI agents.
Personal apps and owning your data
Judging by the chapter list, the episode then moves on to personal software, personal databases and data ownership. The idea is that more and more apps will be built for a single user or a single team rather than for millions. That means vastly more databases, and the question of who owns the data becomes a practical one.
For a business owner this means two things. First, you can keep your own customer and sales data in a structured form without a heavyweight system purchase. Second, the responsibility shifts to you: backups, access rights and data protection are not taken care of just because the app was quick to build.
Agents as database users
Later in the episode the conversation turns to building for personal agents, using your own product in your daily work (dogfooding), and agent-first design and evaluations. It also covers how Supabase has adopted AI inside its own organisation and the lessons learned along the way.
When an agent rather than a human is a service’s most important user, the product has to be predictable, well documented and testable. The same goes for a small company’s own automations. We have also looked at why openness and portability matter in our article on open AI models.
Once you can build a tool yourself, the bigger question is what data it handles and who is responsible for it.
What should founders do now?
- Pick one recurring task, such as tracking quotes or bookings, and try building a small tool of your own for it.
- Before you build, decide where the data will live, who can access it and how it will be backed up.
- Keep a record of what the tool does so that someone else can maintain it too.
- If you let an agent work with your data, limit its permissions and test what happens when information is missing.
The episode closes with the future of coding and how easier software development opens doors to new kinds of entrepreneurship. For a wider view of the startup landscape, see our piece on customer insight as an AI startup’s competitive edge.
Sources
- Y Combinator: What Happens When Everyone Can Build Software (Lightcone, YouTube) — Based on the episode description and timestamped chapter list published by YC, not a verified transcript. Topics include builders vs. traditional developers at 4:34, personal databases at 12:12 and agent-first design at 30:01. The practical suggestions are the author’s own.
Innohub TV
Watch this content also on Innohub TV
We picked clips from Innohub TV that continue the same topic.
