Nestd

Nestd came out of first-hand frustration: finding a rental in the Greater Toronto Area is miserable. Listings are scattered, the good ones vanish in hours, and agents sit in the middle taking a cut. Nestd is an AI-native, direct-to-landlord platform built to run the whole rental loop, from browse to screen to lease to rent collection, without the agent or their commission.

What I built

• A Turborepo monorepo spanning web, API, and shared packages, implementing the full rental loop end to end in development: browse a listing, get AI-screened, sign a lease, pay rent.
• Nine Claude-powered AI services (applicant match scoring, natural-language listing search, listing-description generation, roommate compatibility, and student/newcomer screening), each with a deterministic heuristic fallback so nothing breaks when there’s no API key.
• Screening built for the people the system usually rejects: a built-in guarantor flow with no Canadian credit history required, designed to get international and first-year applicants past automatic rejection.
• AI-assisted Ontario tenancy documents: the RTA Standard Lease plus LTB N1/N9/N11 notices, rent receipts, and condition reports, with rent-guideline-cap checks baked in.
• A verification and trust layer: government-ID and proof-of-ownership checks, signed short-expiry document URLs, and an objective listing-quality grade. Plus Stripe (test mode), Clerk auth, Pusher messaging, Mapbox, and transactional email.

Why it matters

The decision I keep coming back to on Nestd is the fallback rule: every one of the nine AI services has a deterministic version underneath it, so the platform still works with no API key at all. That sounds like over-engineering until you remember a rental platform moves money and signs leases, where “the model was having an off day” isn’t an acceptable failure. And most of the real work wasn’t the AI anyway. It was the unglamorous marketplace problem of verifying listings and earning enough trust that someone will sign a lease through a site they’d never heard of last week.

What it taught me

Framing Nestd as a focused fix for one painful market rather than a thinner clone of the national listings sites forced me to defend why it’s a business, not just a build. Designing the whole rental loop and then having to argue for it as a product taught me more than the code alone would have.