TEMBO

Social media makes it easy to share moments, but harder to keep the experiences behind them connected. We run, cook, travel, and build things with friends, yet those experiences end up as scattered posts or across disconnected platforms Strava, Letterboxd, notes apps, and group chats.

A trip, a friendship, a fitness journey, or a project has its own history, context, and people who are part of it. Together, these experiences tell a fuller story of who we are.

I wanted to explore a social experience where those different parts of life could continue over time, keep their context, and reach the people they matter to.

TEMBO is built for people doing something over time. Each ongoing activity becomes a Storyline, with its own history and audience.

Moments are the updates added along the way.

People can follow the Storylines they care about, rather than a whole account.

I designed Storylines with their own identity, history, lifecycle, and audience, giving Moments context as they build over time.

I designed layered privacy around this structure, where the account sets the widest audience, each Storyline can narrow it, and Moments inherit that choice. These rules extend through the database and media delivery, so privacy is enforced at the system level, not just in the interface.

Posting a Moment takes one tap. Behind it, TEMBO checks the Storyline can still take Moments, screens the photo before anyone sees it, and saves the Moment and its media together, so nothing is ever half-published.

Each photo is then delivered under its Storyline’s privacy rules, and deleting a Moment removes its media too. The complexity stays in the system, so posting stays one tap.

KESHO (opens in a new tab) taught me the cost of overbuilding. With TEMBO, I focused on the foundations of a complete social product, building Storylines, photo Moments, discovery, following, and privacy while leaving more complex features for later.

I also chose a simpler, proven stack of React Native, Expo, TypeScript, and Supabase, allowing me to build for iOS and Android from one codebase. This approach helped me secure App Store approval in five weeks while entering Google Play closed testing at the same time.

I designed and built TEMBO through an AI-native workflow with Claude Code. AI explored and challenged every decision, but nothing was built until I approved it, and nothing shipped until it passed automated tests, security checks, and device testing.

Before release, I tested TEMBO across real devices, core user flows, authentication, privacy, and security. I built automated app and database tests covering permissions, concurrency, failure scenarios, and critical product behavior.

I also load-tested the backend on synthetic data in an isolated environment, and before submission brought in a senior software engineer for an independent review of the app for release certification.

Building a working app wasn’t enough. I had to meet Apple and Google’s requirements around user-generated content, including privacy policies (opens in a new tab), content moderation, reporting, blocking, and account deletion.

For me, onboarding is a launch blocker, so I refined the first-time experience alongside the compliance work. I invited users into Android closed testing, prepared the native builds, and worked through Apple’s review process to secure App Store approval, followed by Google Play approval two weeks later.

As TEMBO moved into production, I established a lean foundation for maintaining and improving it, using GitHub Actions for automated quality checks, Sentry for error monitoring, and PostHog for product analytics.

I deliberately kept the infrastructure simple, focusing on catching regressions, understanding failures, and learning from real usage without introducing unnecessary complexity.