Case study · our own product
Bailar — a global dance-discovery platform, built with the exact method we use for clients.
Full disclosure, up front:
Bailar, Inc. is a sister company under common ownership — not a LaunchOwn client. Bailar is our own product: what the method has already built, shown here with Bailar’s permission. Judge the method by it.
The scale one person can run with this method.
- 179
- countries covered
- ~30,000
- events indexed
- ~11,000
- studios listed
- 30+
- dance styles
- 200+
- locales
- 5
- platforms shipped
iOS · Android · Windows · Amazon Appstore · Apple TV — plus the website, a studios SaaS, and TV builds packaged for Fire TV, Samsung, LG, and Roku. One person, AI-native tooling.
Consumer · mobile
The consumer app.
Consumer · web
The website and directory.
Consumer · ten-foot
The TV app.
Live on Apple TV; TV builds packaged for Fire TV, Samsung, LG, and Roku.
B2B
Bailar for Studios — the B2B SaaS.
A 24-module owner portal: schedules, students, staff, finance, payouts, exports, analytics, marketing tools, and a public page for every studio.
The back end
The part nobody sees — and everybody depends on.
Most of the engineering isn't in the screenshots. This is what keeps a five-platform product running unattended.
The data pipeline
Scrapers and importers feed a staging area with dedupe keys and timezone-integrity rules; a human review gate promotes records to production. ~30,000 events stay fresh without anyone typing them in.
Over-the-air updates
An automated rail publishes fixes to both app binaries with guard hooks and a written rollback runbook — bug fixes reach live users without waiting on app-store review.
A self-hosted build fleet
GitHub Actions routes onto a fleet of owned and flat-rate dedicated machines with documented failover tiers. App-store builds, tests, and deploys run without a per-minute cloud bill.
Crash-to-fix automation
Error monitoring on every surface feeds an automated triage pipeline that classifies new crashes and drives AI-assisted fixes — from stack trace to shipped patch.
Email that never double-sends
Templated transactional sends with an idempotency ledger, inbound webhook processing, and deliverability monitoring.
Locked-down data
Postgres with row-level security across the schema, CI guards that fail any privileged function without an explicit access decision, and 100+ server-side edge functions.
The method
How one person ships all of that.
Owned compute
Builds, deploys, and data pipelines run on a fleet of owned machines — no per-minute cloud build bills, no waiting on someone else’s queue.
Flat-rate AI tooling
AI coding assistants on flat-rate subscriptions do the heavy lifting; a human scopes, reviews, and ships. Cost stays fixed while output compounds.
Agent-run operations
Routine ops — releases, QA passes, content pipelines — are delegated to configured agents with guardrails. The same setup we hand to clients.
See it live