Jason Liu doesn't hide his stake in this argument. He's a small investor in Lovable, the AI coding platform he's publicly urging to overhaul its revenue model. That context matters — his June 2025 blog post, framed as candid advice to the company, is also a pitch for a model that would deepen creator dependency on a platform he holds equity in.
His course business generates roughly $800K a year. He pays $8,000–$9,000 annually in SaaS subscriptions and more than $100K in platform fees to Maven alone, and he spends hours manually connecting Stripe, Podia, Kit, and Zapier to run what should be commodity infrastructure. At $100/month, no platform can justify solving that for him. The math doesn't work.
His proposed fix is a tiered revenue-sharing program he calls the 'Lovable Partners Program.' Creators would opt in to paying 5–30% of their revenue in exchange for white-glove services: one-click Stripe integration with admin tooling, human support that scales with creator earnings, and infrastructure help including migrating off expensive Supabase setups. The platform only gets paid when creators do.
Whether that alignment holds as businesses scale is another matter. A 30% platform cut that feels rational at $10K/month looks different at $500K. Stripe charges 2.9% plus 30 cents per transaction for payments processing alone — a 30% platform take is a different order of magnitude. Lock-in under a revenue-share arrangement also runs deeper than a subscription: switching means renegotiating not just tooling but the economic terms of your business. Liu doesn't address these tradeoffs, and there's no independent voice in his post doing it for him.
The strategic case he finds most compelling is that every manual service the platform delivers — a refund flow setup, a migration, a discount configuration — generates the training data needed to eventually automate that service for all users. Forward-deployed human support, in this framing, is R&D with the client funding it. It's a model familiar from enterprise software: use high-touch early adopters to learn what to scale, then productize it.
Lovable hasn't publicly responded to the post. The pricing tension Liu identifies is real — the cohort of builders shipping functional products through AI without traditional engineering backgrounds is now large enough that flat-fee models built for hobbyists are starting to creak. But the specific solution he's proposing would benefit a platform he invested in, and that's the variable his essay quietly leaves out.