How We Use Google's Gemini AI to Write Smart Descriptions for Every Listing Photo

Most listings on most booking sites have photo alt-text that looks like IMG_3982.jpg. Quietly, that costs hosts in Google Image Search traffic and locks out travelers using screen readers. We’re doing it differently — every photo on Siargao Booking is described in plain English by Google’s Gemini AI within seconds of being uploaded.

Here’s what it looks like in practice, why it matters, and how it works under the hood.

The problem nobody talks about

When a host uploads photos to Airbnb, Booking.com, or most directory sites, the system stores the original filename as the image’s alt-text. So a perfectly shot photo of a sunset villa pool gets described as “DSC_0421.jpg” to anyone who can’t see the image — including:

• Google’s image-search crawler, which uses alt-text as a primary ranking signal.
• Travelers using screen readers (more common than people realize — about 4% of Filipino internet users, and growing).
• Anyone on a slow connection where images haven’t loaded yet.

The result: hundreds of beautiful photos invisible to search engines, and a worse experience for guests who need accessibility features.

What we built

When a host uploads a photo through our listing flow, we send the image URL to Google’s Gemini Vision model with a small prompt asking for a short, factual description. Gemini comes back in about 2–5 seconds with something like:

“Modern Bathroom With Patterned Tile Shower And White Toilet”

“Sunset View Of Infinity Pool Overlooking Coconut Palms”

“Outdoor Shower Surrounded By Tropical Plants”

That text becomes the photo’s alt-text and a sensible filename, replacing the original camera dump. The host doesn’t have to do anything — it’s automatic.

Why this quietly matters for hosts

• Google Image Search: well-described listing photos can drive surprising amounts of traffic. A sunset shot tagged “Sunset Infinity Pool Siargao” can rank for that search and get clicked by travelers planning their trip.
• Page SEO: alt-text contributes to the page’s overall topical relevance. Twenty well-described photos are a much stronger signal than twenty IMG_4001.jpgs.
• Sharing: when someone shares your listing on Facebook, Twitter, or Pinterest, the alt-text often becomes the default caption. Better alt = better social sharing.

Why it matters for guests

• Screen reader users can actually understand what they’re “seeing” on a listing.
• People on slow connections in remote parts of the Philippines see a meaningful description while images load.
• Anyone who reads alt-text out of curiosity (it shows on hover in some browsers) gets context, not gibberish.

Why it matters for the AI search we’re building

This is the bigger picture. We’re working on an AI search assistant that lets you ask “a villa with a pool and a workspace, walking distance to surf” and get useful results. That assistant needs structured, machine-readable descriptions of every photo, every amenity, and every location. By processing photos at upload time instead of at search time, we’re building the data foundation that makes intelligent recommendations possible.

The trust angle

We’re not generating photos with AI. We’re only describing real photos that real hosts have uploaded. The photo of the kitchen is the actual kitchen. Gemini just looks at it and tells everyone what’s in it.

If the AI gets a description wrong, hosts can override it inside the dashboard at any time. We default to AI but defer to human judgment when there’s a conflict.

What this means if you’re thinking of listing

You don’t have to learn anything new or change how you take photos. Upload your shots normally. Within seconds each one is described, indexed, and ready for search. It’s one of the small things we hope adds up to a better directory for everyone.

Ready to list your home? Start here →