Six months ago, I started building Praka — an AI API gateway — in public.
Every week: real numbers, real problems, real decisions. No polish, no PR spin.
Here's what I learned.
What "Building in Public" Actually Means
Not this:
"Excited to share we've hit a major milestone! 🎉"
This:
"We hit 1,000 users this week. Lost 3 of them on day 1 because our onboarding is confusing. Here's what I'm doing about it."
The difference is specificity + honesty. Vague updates help no one. Specific, honest updates build community.
The Unexpected Benefits
1. Your users find you
I've gotten my best customers from technical Twitter threads where I was just... thinking out loud about an architecture decision.
Someone reads "I'm trying to figure out how to route AI API calls based on request complexity" and thinks "wait, that's exactly my problem."
They check out what I'm building. They sign up.
No ad spend. No SEO tricks. Just thinking publicly.
2. Community gives better feedback than surveys
I posted a thread asking "Should I add streaming support or multi-turn conversation memory first?"
Within 24 hours: 47 replies, 12 DMs. Not just votes — explanations. Use cases I hadn't thought of. Edge cases I'd have hit six months later.
That's a free product design session.
3. It actually makes you ship faster
Sounds counterintuitive, right? Doesn't thinking-in-public take time away from building?
Here's what I found: knowing I'm going to post an update on Friday makes me scope work differently. Instead of "I'll finish this big feature next week," it becomes "what can I ship this week that's worth talking about?"
Smaller batches. Faster feedback loops. Classic agile, accidentally enforced by social accountability.
What Didn't Work
Hot takes don't build trust
Early on, I posted some "hot take" style tweets about AI APIs to get engagement. They worked — for a day. Lots of retweets, zero meaningful follows.
The posts that actually converted readers to users? The ones where I shared a specific technical decision and why I made it.
Vanity metrics lose you credibility
"We're growing 20% week-over-week!" sounds impressive until someone does the math and realizes you started from 10 users.
Be specific. "We went from 800 to 960 users this week" is more credible than percentages — especially when the numbers are small.
Daily posting burns you out
I tried posting every day for a month. By week 3, I was manufacturing content. It showed, and engagement dropped.
Now I post meaningful updates when I have something real to say. Weekly is my natural rhythm.
My Actual Weekly Update Format
Here's the template I've settled on:
Week [N] of building [product]:
📈 This week:
• [Specific metric with real number]
• [Feature shipped]
• [Problem encountered]
🤔 Hardest decision this week:
[One specific tradeoff I made and why]
💡 What I learned:
[One concrete, transferable insight]
Next week I'm working on: [specific thing]
---
[link]
The constraint of this format forces me to actually think, not just report.
A Technical Example: Our Model Routing Decision
Here's the kind of thing I post. This is from week 14:
Spent 3 days debating how to handle model routing in Praka.
Option A: Let users specify which model per request (simple, explicit)
Option B: Auto-route based on request complexity (smart, but who decides "complex"?)We went with A. Reason: developers hate magic. They want predictability.
Added a recommendation UI so users can see what we'd suggest, but they always have final say.
Lesson: when in doubt, give developers control.
This kind of post:
- Demonstrates technical thinking
- Shows you value user feedback
- Gives other devs something to agree/disagree with
- Humanizes the product
The Numbers After 6 Months
I'm not going to fake-humble my way through this. Here's real data:
| Metric | Month 1 | Month 6 |
|---|---|---|
| Weekly Twitter impressions | ~400 | ~18,000 |
| Profile visits from build-in-public posts | — | ~1,200/week |
| Users who mentioned finding me via Twitter | — | ~30% |
| Time spent on "marketing" | 0 hrs | 3-4 hrs/week |
The compounding effect is real. Early posts get found months later. A thread I wrote in month 2 still brings in 2-3 signups a week.
Should You Build in Public?
It's not for everyone. You need to be comfortable with:
- Being wrong publicly
- Sharing numbers that might seem small
- Criticism from strangers
- The discipline to be consistent
But if you're an indie developer or small startup trying to grow without a marketing budget, I don't know a better way.
The barrier to building in public is almost zero. A Twitter account. A commitment to honesty. One update a week.
Start before you're ready.
I'm building Praka — one API endpoint for 100+ AI models, 70% cheaper than going direct. If you've been thinking about building in public and want to talk shop, drop a comment below.
Top comments (0)