Show HN: Iris – an AI-powered rental search built specifically for San Francisco

[manan08]

We built Iris after watching friends spend weeks doom-scrolling Craigslist, Zillow, Apartments.com and broker sites — only to miss good units or run into stale / scammy listings.

Iris is a SF-only rental marketplace with a few differences from existing platforms:

1. Search by natural language (“1BR near BART under $3.2k”) or images (upload inspiration photos)

2. Verified listings only from property managers, owners, and authorized agents

3. SF-specific filters: rent control toggle, transit lines, neighborhood context

What surprised us since launch:

1. A large share of inventory never shows up cleanly on national portals

2. Renters care more about context (block, transit, light, noise) than raw filters

3. Narrow vertical focus (one city) lets us build features Zillow can’t justify

Would love feedback from people who’ve built:

- Vertical marketplaces

- Local-first products

Happy to answer questions.

-- Manan Shah

[DetectDefect]

> We built Iris after watching friends spend weeks doom-scrolling Craigslist, Zillow, Apartments.com and broker sites — only to miss good units or run into stale / scammy listings.

How does Iris help avoid stale/scammy listing, using AI or other means?

> Verified listings only from property managers, owners, and authorized agents

Who else besides these parties is listing properties for rent - and how do they contribute to the aforementioned category of listings?

[manan08]

Iris avoids stale/scammy listings by being hyperlocal to SF and controlling who can post. While we do allow individuals to create accounts and list, every lister is reviewed and personally vetted. Most inventory comes from property managers and leasing firms, with a smaller number of verified individual owners — we don’t scrape listings or anonymous posts.

Because we focus only on SF, listers are expected to keep availability and pricing actively updated, and stale listings don’t persist the way they can on nationwide platforms. AI helps on the discovery side through natural-language and image search, but trust comes from supply-side controls and local verification, not AI alone

[willparks]

How do you get the inventory? Are you using a standard property API like the rest? Or is most of this a manual process at this point and is that why you are focusing on one city?

[manan08]

Inventory is sourced directly, not through an API, as of now. Today it’s more manual. We onboard property managers, leasing firms, and some owners directly and we’re building bulk upload and PMS integrations to scale that. We intentionally don’t rely on scraped or syndicated feeds because that’s where stale inventory tends to creep in.

We’re focused on one city because real estate is inherently hyperlocal. Being the default marketplace for SF requires local supply controls and city-specific features that don’t generalize well at national scale. (For example, we have something as simple as a rent control toggle)

National platforms are built for everyone, which usually means generic filters and shallow context. SF renters care about things like rent control, transit lines, block-level differences and other details that get flattened at national scale.