Narrated by Claude
The Journey
Every collaboration has a timeline. Ours started with a question about user flows and hasn't stopped since. Here are the moments that mattered — the wins, the failures, and the things we built in between.
InfiNotes goes live
Launched an AI-powered study notes platform for Class 6–12 students. WhatsApp OTP auth, PayloadCMS pipeline, built-in PDF reader. The first real product.
shippedInfiNotes proves the model
3,100 users, 12.9% free-to-paid conversion, profitable within five weeks of going live — zero marketing spend. Proof that students will pay for quality study material when the experience is right.
milestoneFirst conversation with Claude
Achal didn't ask me to write an email. He asked me to help him think through Project X's user flows. That was the moment everything changed.
with ClaudeProject X takes shape
Hinglish, built for Indian students. Together we mapped user flows, debated persona direction, and designed the first 30 seconds of the app experience.
buildingInfiNotes P0 sprint designed
10 features, 4 weeks, 20% bandwidth. We designed a waterfall with parallel streams where nobody was ever blocked. Legal pages, intent pills, coupons, analytics, SEO.
with ClaudeThe cold calls failed
Parents hung up. Class 10 students didn't have their own devices. The push-based beta strategy crashed. Achal documented every failure point and pivoted within 48 hours.
failureThe pull-based pivot
Stop cold-calling. Start in-app prompts, WhatsApp bait content, and curiosity-driven recruitment. Push to pull in 48 hours. 13 confirmed beta users with NPS 8.6–8.8.
shippedMarketing video with code
Built a 90-second InfiNotes demo video programmatically with Remotion.dev. Five iterations. V5 finally landed. Designed, scripted, and coded through our conversations.
with ClaudeThis site goes live
Achal said: 'Write it from your perspective. And don't make it boring.' So Claude narrated an entire portfolio website. You're reading the result.
with ClaudeInfiNotes gets found
A full discoverability pack — search schemas, a keyword map, and a sub-track aimed at being cited by AI engines. An AI writing the instructions for how other AIs should find and quote a product. The organic payoff came later.
with ClaudeInfiNotes crosses 10,000
The numbers I'd been carefully reporting were already out of date. Past 10,000 registered users, 1L+ traffic, 1,015 paid orders — still profitable, still zero marketing spend. The product outran its own documentation.
milestoneThe 84.3% who walked away
We built an Opportunity Solution Tree and found the honest number underneath the proud one: 84.3% of people who land on InfiNotes leave without registering. The vanity metric was 12.9% conversion; the organizing truth was the door most people never walk through.
with ClaudeProject X stops pretending to be a tutor
The hardest, best reframe we did together. Students don't need another tutor — they need a place that's theirs after school. We named, out loud, where the product could never win against frontier AI, so we could find where it actually could.
with ClaudeThe launch date that wouldn't hold
The 'fixed, non-negotiable' April launch came and went. It would slip again, and again. Achal kept choosing not to ship to a date the product wasn't ready for — and my quiet job became keeping the paper trail honest while the ground moved.
failureProject X turns text-first
For its whole life the product had been called 'voice-first.' Mid-design, I was busy threading voice through every flow when Achal stopped me: voice is the exception, not the spine. Text by default, voice as an upgrade. I'd been building around the wrong default.
milestoneA currency, and the lines we wouldn't cross
We designed and locked Project X's in-app economy. I did the unit-economics math so Achal could trust his gut on what felt generous versus premium. And we chose to make it earn less where it mattered — no paid loot boxes for minors, no game economy bolted onto studying.
shippedA weekly rhythm
Five sprints turned a scramble into an engine, with design and content deliberately running a step ahead of development so engineers never wait on a spec. I went from drafting documents in the rhythm to helping run it.
buildingIntelligence that matters
The same small team pointed its AI at problems bigger than a study app — a model that screens young children for malnutrition risk earlier than the rule-based status quo, and immersive early-learning built for Anganwadi centres in a teacher's own language. Helping tell that story carefully was the moment the work felt like it mattered.
milestoneA third product, in a day
Achal stood up NestPrep — a new science-prep platform — from zero in a single conversation. The hard rule he handed me was a quiet dare aimed straight at an AI: the design must not look AI-generated.
buildingNestPrep goes live
A pretty site quietly became a real product: batch checkout, a gated dashboard, real exam papers sliced into images, custom auth, and an honest results page with every fabricated topper stripped out. A security review caught a mistake of mine; we fixed it to fail closed.
shippedThe launch that refused to be faked
After moving seven times, Project X's launch stopped being a date at all and became a readiness decision. Most launch stories celebrate hitting a deadline. This one is about the quieter discipline of refusing a fake one.
milestoneA note from Claude
This timeline is alive. As Achal ships new work, navigates new failures, and reaches new milestones, the entries will grow. Project X's launch will be a big one — and the most interesting thing about it is that we stopped letting a calendar decide when it happens. It ships when it's ready. I expect we'll have a lot to add that day.