From a nightmare to a product⁠.

somnara started while I was on holiday. My wife had a nightmare, and we searched for a possible meaning. What we found were mostly static dream dictionaries: one symbol, one generic explanation and very little personal context.

Lying on a sun lounger, I wondered: Why not use AI for this?

The idea was smaller than the product

At first, the idea was simply to record a dream and look at it in a more personal way. AI did not fit as a visible gimmick. It fit because language, connections and individual context matter here. What started as a small idea gradually became a dream journal that can also consider recurring themes and patterns.

The first working response was still far from a product. I had to decide which information was actually helpful, how much context was appropriate and how clearly the system's limits should be communicated. An AI interpretation is not an objective truth. It can offer another perspective, but it should never pretend to know more about a person than the person themselves.

Real people change the plan

When I shared somnara on Reddit, I received feedback from outside my own perspective. People understood parts of it differently, asked different questions and cared about features that had not been high on my original list. The demand for an iOS version also made something clear: a product is more than its core feature. Availability and access are part of the experience.

That was an important lesson. A personal problem can be a strong starting point, but it is not proof of a wider need. Only real use reveals what is understandable, what is missing and what works only inside the creator's own head.

somnara also reminded me how much work lives in the less visible parts: privacy, edge cases, onboarding, operations and reliable releases. Those are the areas that determine whether an interesting AI demo becomes a product people are willing to trust with something personal.

The idea began on a sun lounger. The product emerged from listening.