LawnscapeGuru

The premise

Lawn care advice is abundant and almost always wrong for you, because it’s generic. When to water, when to fertilize, when to overseed, and whether to do anything at all this week depend on your grass type, your soil, your climate zone, and what the weather is actually doing right now.

LawnscapeGuru takes a questionnaire about the property, resolves the location to a hardiness zone, pulls current and forecast weather, and generates guidance for this specific lawn this specific week.

What makes it more than a prompt wrapper

The weather integration is what makes the output non-generic. “Fertilize in early spring” is a search result. “Hold off on fertilizing — you have heavy rain coming in three days and it’ll wash off” requires knowing where the user is and what’s about to happen there. The model reasons over real environmental inputs rather than restating general advice.

A ZIP-to-hardiness-zone lookup handles the climate context, and a weekly digest cron job re-evaluates conditions and sends updated guidance — which turns a one-time recommendation into something ongoing, and is the part that would justify a subscription.

Implementation notes

• Provider abstraction with Grok as primary and optional OpenAI fallback, behind an interface so the model choice isn’t load-bearing
• Subscription gating with a free generation limit before full access is required, enforced server-side
• Postgres via Prisma modeling subscriptions, recommendations, saved locations, payments, newsletter subscribers and affiliate products
• 22 API routes covering recommendations, weather, Stripe webhooks, subscription management, newsletter and the digest cron
• Contentful for the editorial content, and Resend for email
• Test coverage with Vitest — partial, but present

Honest assessment

Live and functional, with the full commercial apparatus in place: auth, payments, metering, scheduled jobs. I don’t have adoption numbers worth quoting, and I’d rather not invent any.

What it demonstrates is that I can take a personalization idea through the whole stack — external data integration, LLM reasoning over structured inputs, subscription billing, scheduled background work — and get it deployed.