The Final Isn't in the Journal

By Georg Malahov

I was flipping through this summer's journal and noticed something odd. The whole Referity race is logged day by day: the January coworking, the application, "we're in", the pitches, the top ten on my birthday. And the final itself isn't in the journal — not a single voice note. Partly it's mechanics: judging by the tasks from those days, I dictated the wrap-up not into the diary but straight into Referity, as feedback inside the project. And partly it's more honest than that: I was properly upset and dropped everything at once, in the heat of it. For a while I didn't record anything at all, not even ordinary notes.

June 26, inside Referity: “Leave feedback on the project” — urgent, due today. The diary stayed silent, but the project recorded it

Now that the dust has settled, I can tell how it went. We prepared for the final honestly — just for the wrong thing. Most of the time went into polishing the product: building features, fixing bugs, wanting everything to work. That would have made sense if we'd shown a live demo — but we ran out of time for one, and what remained was a standard deck along the pitch structure: what it is, who it's for, what the market is, what the prospects are. We presented without much energy: either nerves got to us or we hadn't had time to rehearse. It came out boring. And then a question flew in from the jury, standard and tricky at the same time: why a separate agent, if inside ChatGPT there's already chat, voice, search, context, and connections to external services? We couldn't find an answer. Possibly that was the key moment of the whole final.

We didn't take a prize. That we came fifth I learned only later, from a conversation with the accelerator's organizer: the jury debated for a while over which place to give us, and there was a real shot at the last prize line — in the end we slid to fifth, right behind the prize-winning four. After the last meeting I skipped, fifth is closer to a lot than to a little. We got a paper finalist diploma — a consolation prize you can put on a shelf; somewhere on a shelf is where mine is. And we didn't part in silence: the team met, then we all got on a call with the organizer and went through what worked and what didn't. The conclusions from that call I still remember, without any notes. Marketing decides as much as the product — and we'd been building only the product the whole way. The presentation on stage matters more than the app itself: the jury sees the performance, not the months of code. And it needs to be prepared like a separate product, in advance — not the night before the deadline, the way we did it every single time.

What didn't happen after the debrief was the next step. The conclusions were there, but the next working meeting never got scheduled. The second co-founder and I keep working on other things; the first went off into family matters and building a house. The team didn't fall apart over a quarrel — the project simply ended, and a new one didn't begin. In hindsight it's visible: we had assembled around the application and the accelerator's deadlines, not around the product. The deadlines ran out — and the calendar went empty.

July went into being torn between everything at once: promoting the bot, building something new, where to get money. Seventeen minutes into the journal — and not a single decision. Then I cleared the decks and picked a focus: a conversation with my partner at the day job on August 1 put everything in place — right now the main thing is there, not in the pet projects. I agreed and felt relief. That relief was the real answer about Referity — I just didn't hear it right away.

Now, by mid-September, I can lay my own layer on top of the team's debrief. When people asked who this product was for, my honest feeling was: for everyone. That is the wrong answer. We stuffed everything into Referity: projects, subprojects, themes, topics, a task tracker with tasks assigned to people, team access, notifications, storing photos and videos as a second memory — a Frankenstein of a tool, teamwork, and a life diary. And it didn't solve any one concrete pain — it was an interesting product that test-drove modern tech. Meanwhile the prize places went to projects that had no product at all: they had an idea and a clear answer about who it was for. We probably had the most built-out product of the whole batch — with a website, with an app you could install and use — and we couldn't get across to anyone why it was needed. Getting that far with the product and doing that badly at finding an audience isn't a jury remark, it's my own stable pattern: building I can do and love, selling — no. Plus the pace: a month's task in one week — on evenings after work and family, that holds for a month or two, but not for half a year.

The structure of those weeks as it was: feedback on the app, feedback for Startbahn27

What's left, if I count in something other than money. The journal is alive: I dictate into it every day, and out of this whole story it's the only thing that works with zero effort — turn on the microphone, talk, everything saves and sorts itself. One simple, perfectly clear job. Looks like that's what a product is — while Referity wanted to be everything at once. What's left is experience that didn't exist in January: the first application, the first team, the first pitches in German, an accelerator from start to finish — with a fifth place. And the habit of paying a hundred euros a month for development speed — that one will outlive any project.

So the next run will be different. I don't want to build product for product's sake anymore, features for features' sake. Without users a product is pointless — that's the understanding I left the accelerator with. Not building anything new and big — first learning to take what's already built all the way to money. I signed up for a marketing course: I want to figure out my audience with people who've done this with their own hands, instead of re-deriving everything through my own bruises. I'm not burying the Referity idea: the database, the parsing, and the voice input haven't gone anywhere, and if I ever come back — I'll come back to working code.

A day assembled from voice notes — with quote-links to the originals. This is what it was all built for

Testing all of this is simple: the next project has to get where Referity never did — to the first money from a stranger. Not to a pitch, not to a fifth place — to a payment.

And the hole in the journal can stay. Back then I dictated the final's wrap-up not into the diary but into Referity itself — so the project recorded its own ending.