I'm a solo developer in New Delhi. For the last seven weeks I've been building Align, with AI coding agents doing most of the typing. Here's what that looks like on paper:
- 1,000+ commits (around 190 in a single week)
- ~375K lines of code across 14 apps
- AI meeting rooms, docs, chat, a task tracker with two-way Jira/Linear sync, a Chrome extension, a Mac menu-bar app, an MCP server, a status page, an admin dashboard
And here's what it looks like in reality:
- 2 outside teams use it. Neither pays.
- ~195 landing page visitors in 15 days
- 78% of them saw one page and left
- 0 clicked the main "Start a room" button
I'm writing this down because I think a lot of people building with agents right now are about to hit the same wall, and because saying it in public makes it harder for me to keep doing the wrong thing.
What went wrong
1. Agents made building free, so I never had to choose.
Before agents, building a feature cost a week, so you had to think hard about which one mattered. When it costs an afternoon, you just build all of them. Every time a user didn't appear, my reflex was "it must be missing X." So I shipped X. Then Y. Then a whole docs app.
None of it was the problem. Nobody was missing X. Nobody knew Align existed.
2. I built a suite before anyone wanted one product.
My landing page tried to explain six apps. A stranger has about five seconds to work out what something is. "AI decision infrastructure for your whole team" means nothing in five seconds. The 78% bounce is the page telling me that.
3. I never launched.
I wrote a go-to-market plan with Show HN and Product Hunt dates on it. Those dates came and went while I shipped more features. LinkedIn posts reach my own network. SEO brings about 10 visits. Building felt like progress, and launching felt like risk.
4. I picked the most crowded category there is.
"AI meeting notes" means competing with Otter, Fireflies, Fathom and Granola, and all of them have free plans. "Better notes" is not a reason for anyone to switch.
What I'm changing
One product. The Chrome extension, Align for Calls. It's the easiest thing to try, and it has the one real difference from everything else: no bot joins your call. It reads the live captions Google Meet, Teams web and Zoom web already show. Nothing records audio. During the call it tracks decisions, owners and open questions. After the call you review a wrap-up, email it to the attendees, and the action items become tasks you can sync to Jira or Linear.
Everything else stays live but frozen. No new features until 20 teams use the extension every week.
One page. The landing page will explain the extension and nothing else.
Talking instead of shipping. Every comment on this post gets a reply from me, and anyone who tries it gets a message asking what broke.
What I'd love from you
- If you've been through this: what got you your first 20 users? Specifics, please. Which community, which message.
- If you hate meeting bots, try it and tell me what's wrong with it: https://meet-align.com/extension
I'll post the real numbers again in 30 days, good or bad.
Top comments (2)
Man, this one stung a little because I'm living the smaller version of it. I'm a school teacher in Pakistan, not a developer — I've been building free AI tools with AI help (a voice demo, a form builder, stuff like that). And yeah, building is the easy part now. The wall is exactly what you described: nobody knows the thing exists.
One small thing that's actually worked for me: just writing about the build here on dev.to. My first article pulled 48 views in a day. Tiny numbers, I know — but those are 48 people who came to read, not bounce. Posting the work in public, even messy, has beaten every "one more feature" impulse so far.
Really rooting for the one-product bet. Following for the 30-day update.
0 clicked the main button is the number that makes the whole post-mortem honest.