The Process Problem
A new listing hits the MLS feed with a full set of photos, price, beds, baths, and agent remarks — and then, for most agents, nothing happens with that data beyond the listing itself. Turning it into a short video or social post for Instagram or Facebook means someone manually downloading photos, writing captions, and either editing a video by hand or skipping video content for that listing entirely. Multiply that by every new listing in a month and most of it simply doesn't get made.
Automation Approach
The workflow watches the MLS feed (via RESO Web API or RETS, depending on what the MLS supports) for new or updated listings and pulls the structured fields the moment one appears: address, price, beds/baths, square footage, photos, and agent remarks. Photos are ordered — hero shot first, usually the best exterior or interior room — and validated for minimum resolution. Key facts are formatted into on-screen text overlays, and an AI step drafts a caption variant from the agent remarks for the social post. The photos are then assembled into a vertical 15-30 second video with a branded end card, alongside a static image-carousel variant from the same assets, and both are queued to the agent's or brokerage's social accounts on a set posting schedule.
Why This Approach
Feed ingestion and photo ordering are deliberately rule-based — there's no judgment call in detecting a new listing or sequencing photos by a fixed heuristic, so a plain automation step is faster and more predictable than an AI one. The caption draft is where AI genuinely earns its place, because turning a few lines of agent remarks into a social-ready caption is a writing task, not a lookup. This workflow does not replace a photographer; it starts from whatever photos already exist in the MLS listing and automates everything downstream of that.