Bonzai’s first week changed more than the prototype

In less than a week, I took Bonzai from a concept to a voice-first publishing prototype that I thought was working quite well. I ran a lot of agents in parallel, had them work through features as I saw them, pushed the prototype live and put it in front of my co-founder. The speed felt extraordinary.

The initial idea was closer to a publication. An AI journalist would interview people and have a fairly strong editorial opinion about what went out. My co-founder was engaged and gave me good feedback as we worked through the prototype. But feedback was not the whole reason the idea changed.

We are hoping to raise funding, and we want Bonzai to be positioned as a publishing platform, not as a publication. I think that argument alone is enough. It is a decision about how we present what we are building, as well as the direction of the product itself.

We have moved toward something less opinionated and more centered on user-generated content. Medium, Substack and social platforms are points of comparison for what I mean: a simple, appealing place for people to bring their own material, rather than a publication whose AI journalist has a strong say in what gets published. That is the direction we are taking, not a claim that people are already using Bonzai that way.

Running agents in parallel helped me put a working prototype in front of my co-founder quickly. It did not, by itself, settle what Bonzai should be. That first week brought a distinction into focus: Were we building a publication with an editorial point of view, or a platform where the content would come from its users?