// how i think about building
BUILDING
PHILOSOPHY.
Not frameworks. Not startup advice. Just the real lessons from building ExplainMate and Plintrix — including the things that failed.
Traffic without retention is noise.
When I launched ExplainMate, a reel went viral. Thousands of people landed on the site in a short window. I was watching the analytics in real time — it felt like validation. Then I checked how many actually used the product. Almost none. Sign-in was broken on some devices. The moment they hit friction, they left. And unlike paid traffic, viral traffic doesn't come back. That loss taught me more than any startup book: the job isn't to get people to your product. The job is to make sure they don't leave.
→ Retention > Traffic. Every time.
A broken product at the wrong moment destroys more than zero users.
The ExplainMate viral moment came before the product was ready for it. The sign-in issue wasn't a major engineering failure — it was just not tested across all device types. But when thousands of people hit that friction point at once, the opportunity was gone. They didn't bookmark it to come back. They just left. Momentum is fragile. You can survive a slow launch. You can't always survive a broken first impression at scale.
→ Ship complete. Not feature-complete. Complete.
Execution is the only thing that separates ideas from products.
Everyone who's ever heard of ExplainMate or Plintrix has said 'I had an idea like that.' I believe them. Ideas are genuinely common. What's rare is the willingness to spend a week debugging auth flows, rewriting onboarding three times, shipping at midnight, and doing it again the next day. Execution isn't a skill. It's a decision — to keep going after the interesting part is over.
→ Ideas are free. Execution is the actual work.
AI should solve problems. Not just generate text.
The AI space is full of products that are technically impressive and practically useless. They generate. They summarize. They rephrase. But they don't solve a specific problem for a specific person in a way that makes their life meaningfully better. ExplainMate exists because I believed AI could teach — not just answer. Plintrix exists because I believed AI could help people actually execute on their goals, not just get advice. The question I ask before building anything: what does this actually do for the person using it?
→ AI that doesn't solve a specific problem isn't a product. It's a demo.
User feedback is the only reliable source of truth.
I've been wrong about what users want more times than I've been right. I built features I was certain would be used — they weren't. I ignored edge cases I thought were minor — users hit them constantly. The only thing that consistently improved both products was talking to users and watching how they actually used the product. Not how I imagined they would. How they actually did.
→ Your assumptions about your product are not facts. User behavior is.
Real products beat portfolio projects.
I could have built a dozen impressive-looking demos. But demos don't have users. They don't have real feedback. They don't break in ways that teach you anything. ExplainMate and Plintrix exist because I wanted to build something that people would actually use — even if it meant shipping something imperfect and having to fix it in public. The decision to build real products instead of portfolio pieces is the single most important decision I've made as a builder.
→ If nobody uses it, it doesn't matter how impressive it looks.
These lessons are built into every product Avenoric ships.