Building in public has become a genre. The format is predictable: share a revenue screenshot, add a lesson learned, post on Twitter, collect engagement. The insights are usually tidy. The metrics always go up. The failures are retrospective and wrapped in redemption narratives. It is content marketing disguised as transparency.
I want to try something different. I want to share what building a product actually looks like, not the curated version. The version where you spend three weeks on a feature that nobody uses. Where you wake up at 2am worried about a billing bug. Where your co-founder disagrees with you about something fundamental and you cannot resolve it with a whiteboard session. Where the metrics go down and you do not know why. That version.
The reason honest building-in-public matters is that the curated version sets false expectations. Founders see other founders' polished narratives and think they are doing something wrong. Why is my journey so much messier? Why do I not have clean lessons learned? Why does nothing follow the neat narrative arc of problem-insight-solution-growth? The answer is that nobody's journey is clean. The clean versions are edited after the fact.
Here is what the last month of building mani actually looked like. Week 1: we shipped the daily queue feature. I was excited. It worked well in testing. In production, the overnight generation ran out of memory on 3 of 12 brands because their DNA profiles were larger than expected. The morning queue showed 6 empty slots where creative should have been. Fixing the memory issue took 4 days, during which I apologized to beta users every morning and manually generated creative to fill the gaps.
Week 2: a beta user told me the brand colors were wrong. I checked. They were right. Our color extraction algorithm was pulling the CSS custom properties, but this user's site used inline styles exclusively. The extractor missed every color. Their Brand DNA had our default palette instead of theirs. Fixing the extractor to handle inline styles took 2 days and broke another extractor that relied on the old output format. Fixing the break took another day.
Week 3: we launched the performance dashboard. It showed the wrong metrics for the first 48 hours because the timezone handling was off. Every user saw yesterday's data labeled as today's data. Nobody noticed except me, because the data was close enough to be plausible. I fixed it and told nobody. This is the kind of bug that building-in-public accounts never mention because it is embarrassing and undramatic.
Week 4: I reviewed the NPS scores from the first 30 days. Average was 6.8. Not terrible. Not good. The detractors all cited the same issue: the creative was "too samey." Different angles, different formats, but a recognizable AI-generated aesthetic that felt repetitive after seeing 50 pieces. This is the hardest feedback to address because it requires improving the generation engine, which is a multi-month effort, not a weekend fix.
None of these stories has a clean lesson. The memory issue taught me to test with production-size data, but I knew that already. The color extraction bug taught me to handle edge cases, but every developer knows that. The timezone bug taught me nothing new. The NPS feedback taught me that my generation engine needs fundamental improvement, which I already suspected. The honest truth is that most weeks do not produce insights. They produce small fixes to small problems, and the aggregate of those fixes is what eventually becomes a decent product.
The financial transparency that building-in-public celebrates is also more complex than it appears. Our MRR is $1,200. That sounds like traction. But our server costs are $800, our API costs are $300, and my time opportunity cost (what I could earn consulting) is $15K/month. By any honest accounting, we are burning money. The $1,200 MRR is a metric, not a business. Sharing it without the cost context is misleading, which is what most build-in-public accounts do.
I am going to keep sharing honestly, not because it makes me look good (it does not), but because the curated version of entrepreneurship is doing real damage. Founders are making decisions based on false narratives. They are comparing their messy reality to polished fiction. They are feeling inadequate about outcomes that are actually normal. Someone needs to show what normal looks like. This is my attempt.
Mani's public metrics, pivots, and mistakes are published on our site. Not the curated version. The real version. Because if we are building a tool that promises honest brand representation, the least we can do is represent our own brand honestly.
The emotional toll of honest building-in-public deserves mention. When you share only wins, the engagement is positive and affirming. When you share failures and messy weeks, the engagement is mixed. Some people appreciate the honesty. Others judge you for the mistakes. Others offer unsolicited advice that reveals they did not read the post carefully. The emotional cost of transparency is real, and it is part of why most founders choose the curated version. I choose the honest version because the curated version was making me feel like a fraud. At least with honest sharing, the discomfort is external rather than internal.