Last week, Plainline started turning from an idea about "text in, presentation out" into something closer to a product.
Before that, most of the work was built around one question: can a model create a decent presentation with a stable layout from text at all? This week, it became clearer that this approach has limits. So I changed direction a bit and started working on my own render engine instead of simply using AI for that part.
I marked up around 5,000 slides for it, kept researching the "text -> presentation" pipeline, and started building the foundation of the editor. Not just a preview screen, but a system with slide cards, editing forms, navigation, and so on. By the way, during all of this, I almost never opened Figma, as sad as that may sound. Or maybe not sad. I still haven't decided.
In parallel, I kept moving Leaflo forward.
I made the FAQ section in the app, fixed a couple of bugs, went through the guide questions again, and finally put together my marketing and product analytics dashboards. Yay.
The main conclusion of the week is simple: when you build an AI product, it is easy to move too quickly into the magic of the model. But the product does not start where the model generates something. It starts where a person gets a clear way to control the result.
That is what I was trying to find this week.
Next week, I want to bring Plainline closer to one complete flow: paste text, structure it, preview it, edit it, and get a presentation in one scenario.
Let's see where it starts to break.
