Blog
Developer journal, tools, and insights.
The WhatsApp Reply That Reshaped the Product
I sent a SportsSync reel to a cycling creator for feedback. His first reply: 'the Insta360 Go Ultra already does this natively.' What followed was a competitive research rabbit hole, a near-panic moment, and one sentence that clarified who the product is actually for.
What 13 Episodes of Building in Public Taught Me
Thirteen episodes, seven months of development, one idea that survived mostly intact, and a lot of assumptions that didn't. Before moving on to what's next, a honest look at what this series got right, what it got wrong, and what I'd do differently.
First User Research: What a Real Cyclist Taught Me About My Own Product
I sat down with Santi, a 27-year-old cyclist who went from 94kg to 74kg through cycling, and watched him use SportsSync for the first time. His feedback revealed value I hadn't considered and gaps I'd overlooked — including the entire vertical video format.
From Lovable to Claude Code: How I Turned a Prototype Into a Real Application
After three months away, I migrated SportsSync from Lovable to local development with Claude Code. The result: 537 tests, CI/CD pipelines, semantic releases, and a private dashboard with Google authentication. Here's what changed and why.
Designing the Video Processing Backend: A Conversation With a Senior Engineer
I sat down with Tirth, a full-stack engineer who builds MVPs for startups, to design the backend architecture for SportsSync's video rendering pipeline. EC2, S3, SQS, FFmpeg, and the path from single-server MVP to scalable infrastructure.
When Two YouTube Players Can't Coexist: Rethinking the Product Architecture
I spent a session fighting YouTube's iframe API, discovered you can't have two synchronized video players on one page, and realized the constraint actually led to a better product architecture — a separate demo page that becomes the future private dashboard.
The Demo Is Live: From Concept to Working Product on the Landing Page
After nine episodes, the telemetry overlay runs live on the SportsSync landing page. Visitors can see real cycling data synchronized with video. Here's what it took to get here, what's next, and why I'm pausing development to talk to real users.
Polishing the Overlay: From Functional to Professional
The telemetry sync works. Now it needs to look good. I spent a session refining the CSS gauges, fixing cadence and power interpolation, and planning the transition from demo to product — including user research with real cyclists.
The GPS Data Gap: Why My Telemetry Was Out of Sync
The telemetry overlay looked perfect at the sync point but drifted everywhere else. After digging into the raw GPX file, I found the culprit: GPS devices silently drop data points when you stop moving. Here's how I found and fixed it.
Can Lovable Build an MVP? Testing the Limits of AI-Assisted Development
I asked Lovable to build a working GPS telemetry overlay synchronized with a YouTube video — entirely client-side. The result was 80% of the way there in 30 minutes, but the last 20% revealed where AI tools hit their ceiling.
Connecting Supabase to a Waitlist Form and Killing My Darlings
I connected a Supabase database to the SportsSync landing page in minutes, removed sections I loved but didn't need, and learned that a landing page for an MVP needs to be ruthlessly simple.
Iterating a Landing Page with AI: The Before/After Slider That Changed Everything
I spent an hour with Lovable refining the SportsSync landing page — fighting image ratios, removing clutter, and discovering that a simple before/after slider communicates the product better than any copy could.
From AI Prompts to Brand Identity and a Live Landing Page
I used Claude to generate briefs for a Fiverr designer, bought a domain, and built a landing page with Lovable in one session. The logo cost €35, the domain €28, and the landing took an hour. Here's what worked and what didn't.
Using Claude as a VC, User Persona, and Product Manager
I asked Claude to role-play as an investor, a potential customer, and a product manager for my SaaS idea. The results were surprisingly useful — and revealed things I hadn't considered.
How I Used AI to Validate a SaaS Idea in One Session
Before writing a single line of code, I spent 90 minutes with Claude validating a SaaS idea for GPS telemetry video overlays — competitive analysis, SWOT, PRD, and MVP scope. Here's what happened.