Time to stop thinking like a developer

By Georg Malahov

Went on a call as part of the product-launch course and got a fairly unpleasant but, I think, useful kick. The guys looked at my browser transcriber — the extension that sits in the sidebar, listens to a call, and writes out the transcript — and said the MVP works and the market is alive. And then came the part I basically already knew but had been carefully not noticing.

This is what the product already looked like back then — a sidebar with recording, a live transcript, and translation. A frame from the Chrome Store listing; it hasn't changed since launch.

That I've built a pile of stuff. The transcriber, a notes journal, a couple more half-finished things — a whole spread of solutions, something for every taste. And the one thing that actually matters, I'm not doing. I don't put a label on it, don't put the product on a shelf, don't run ads, don't tell anyone about it. Which means I'm still thinking like a developer who enjoys digging around in code, not like someone who wants to get paid for it. It was said to my face, and I had nothing to counter with.

So this morning, October 27, my colleague and I spent only forty minutes on the day job — and I got to sit down and do something unfamiliar: market research. First I pulled competitors' keywords straight out of Google Ads by their URLs, then went to see who's running what in the Google Ads Library, on Facebook, on TikTok. Something curious turned up: the competitors on Google and the ones on Facebook barely overlap, as if these were two separate sets of players. Found a few more companies and products that way that I hadn't known about before.

Meanwhile I got distracted, as usual, by the technology itself — couldn't help it. Checked whether audio can be recognized right in the browser, on-device, through Whisper, without sending the sound off to any server. And it works. Quality is weaker for Russian, but overall it runs. And then a thought came together that I liked myself: do everything the competitors do, but stand out on privacy. Build the same service entirely on local in-browser models, so the audio never leaks anywhere. It seems like for part of the audience — the people who care about their conversations — that could be an argument. Though for now it's only a guess — I still need to look into whether people write about concerns like that at all, or whether I made it up.

Pricing I sketched out roughly like this: ten euros a month for a thousand minutes, or a hundred euros a year, with part of the unused balance carrying over. The funnel in my head is the standard one — ads, landing page, extension install, free trial, form, payment. On paper it's neat. On paper it's always neat, of course.

Out of all this I think I'm starting to understand my main problem, and it isn't technical. I know how to build and I love building, and all of this — research, the label, the ads — is foreign territory for me and slightly off-putting. But since it was said to my face, I have to push myself through it. So the near-term plan: write the product description out in full, do the marketing homework, get the landing page's front page ready. And only then go back to the code, or I'll get carried away again and spend a week in the editor grooming things that already work.

Whether I've moved off dead center or not will show in one sign, an inconvenient one for me: the landing page is done and the ads are running — not "the product got a little better again." If in a couple of weeks I have at least some kind of ads running to a live landing page, it means I've finally stopped hiding in the code. And if I find myself refactoring again — it means I haven't, and I'll have to admit it to myself.

For now I'm off to write the marketing homework — which for me is a feat in itself.