VELVIOO Back to velvioo.com

Notes from the studio

A Founder's Note on Building UnderSelf

Albert Manukyan ·

A closed laptop, an open notebook with a pen and a cup of coffee on a wooden desk at night, city lights through the window

Most of what we do at Velvioo is for people who run fleets. Bikes, scooters, the app a rider taps and the dashboard an operator watches. So it surprises people when they find out the other thing I've been building is an app for dreams. This note is the plain version of why.

It started with forgetting

I forget my dreams by breakfast. For years that was just a fact about me, the way some people can't hold on to names. I'd wake up with something vivid still in my head, a place, a person, a feeling that didn't belong to the day, and by the time the coffee was made it was gone.

At some point I got curious about what I was losing, and not in a mystical way. I just wanted to know what my own head was doing for a third of my life, and I couldn't keep hold of it long enough to find out. Typing on a phone didn't work; the second I unlocked the screen I was awake in the ordinary way and the dream was already leaving. What did work was talking. Say it out loud, half asleep, before anything else.

That was the first version of UnderSelf: a button and a microphone. I recorded the first dream in it in October 2025. I built it for one user, me, and it stayed that way for a while.

What it does today

The core hasn't changed. You wake up, you speak the dream the moment you can, and it's saved before it has a chance to slip. That part has to be fast and forgiving, because nobody is at their best at six in the morning.

Then it tells you what it makes of the dream, and not as a dictionary entry where a river means one thing and a staircase means another. It works from what you've already told it: your earlier dreams, the people and places that keep showing up, what's been on your mind lately. The question it tries to answer isn't "what does this symbol mean" but "what is this dream doing in your life right now." It asks more than it declares, which is how I wanted it.

Over weeks, the entries add up into something. You can see which themes return, when they cluster, and what was going on around those nights. In mine it is water: an ocean right beside a city, a wave coming over the buildings, diving for something dropped in bright water. It turns up in about one dream in eleven. The thirtieth dream is worth more than the first, which is the opposite of how most apps age.

The newest piece is the one I'm most pleased with. The assistant you already use can now talk to your journal. Connect Claude, ChatGPT or Grok through one address, and you can tell it a dream over coffee, ask what you dreamt about last month, or ask what keeps coming back. The journal stays yours; the assistant just gets a door. If you want to see any of this, it's all on UnderSelf, including a free web version you can try before installing anything.

What the two products taught each other

A fleet platform and a dream journal share one hard constraint: both are used by people who are barely paying attention.

A rider unlocking a scooter is standing on a curb, late, with one hand free. A person recording a dream is lying in the dark with their eyes half open. Neither will read instructions or forgive a second screen. Six years on ANIV, our fleet platform, taught me to count taps and remove them, and that discipline went straight into the dream app.

The lesson ran the other way too. Building the journal taught me that once the fast part is done, people want something worth coming back to. A year of dreams should show you a pattern, and a month of rides should tell an operator a story, not just fill a table. That thinking is now shaping how the platform works on the fleet side as well.

An invitation

If you forget your dreams by breakfast too, try it for a week. Speak one dream a morning, even three sentences, and see what comes back. And if you're here because you're evaluating what we build for fleets, this is part of the answer: we care about the first five seconds of a product and about what it's worth on day thirty.

Albert