Homesly: property intelligence with room-by-room scans
Homesly.ai turns a phone video of each room into an inspection record. It flags damage, matches a move-out walkthrough against the move-in one, works out what the tenant owes and sends the repair items to a vendor. MCCSoft built the application and the vision pipeline behind it, and runs both. It is sold to property managers, landlords and maintenance teams.
- 4steps from walkthrough to work order
- 3 of 5scoring runs must agree on a finding
- $2 to $3per unit per month, every feature included
What it does
A manager records a short video of each room with a phone. The AI reviews the footage, flags damage, notes where it is and rates how severe it is. At move-out, Homesly pairs the new footage with the move-in footage from the same angles and separates new damage from what was already there. Borderline matches are kept and flagged, never dropped.
Findings become line items. The system subtracts tenant-caused damage from the deposit and produces a branded PDF with side-by-side photos for each claim. Selected items go to a contractor as one work order; the vendor accepts or declines from a secure email link with no login. A tenant can run their own move-in or move-out walkthrough from a link, and an upload resumes after a dropped connection.
Pricing is per unit per month: $3.00 for the first ten units, falling to $2.00 above 500. Every plan has every feature, and the trial is 14 days with no card.
What MCCSoft built
Version 1.4.0, released 16 October 2026, is a Lucee CFML application on Tomcat with PostgreSQL behind it. The front end is a single-page app in plain JavaScript. Billing runs through Stripe.
The inspection pipeline cuts each room's video into frame sheets. A hosted vision model reviews them in a triage pass and then five scoring runs, and a finding is kept only when at least three of the five runs agree on it. The prompts carry no example values, because a sample finding in a prompt had been copied into real reports. MCCSoft measured that in October 2026 on 61 production scans and removed every example. A footage check flags video that looks edited, composited or screen-recorded; it is a notice on the report, never a block. PDFs are rendered by headless Chrome in two to six seconds.
What it runs on
Production runs on the same dedicated server as MCCSoft's other public sites. Customer data lives in a PostgreSQL cluster with a leader and a streaming replica on separate hardware, with a vector store beside it. The vision inference service runs on a GPU server MCCSoft owns, on its own network.
Three Homesly components sit on the public status board at status.mccsoft.io: the site, the customer data and the intelligence engine. On 18 October 2026 the board showed 100 percent for the site and the engine and 99.99 percent for the data over the previous 90 days.
What it shows about MCCSoft
Homesly is an AI product carried from the first scan to paying customers: a vision pipeline, deposit arithmetic, PDF output, a vendor workflow, billing and a public uptime record. When the model copied a sample finding into real reports, MCCSoft found it by measurement and fixed it in the prompts.
Have a process a camera could document?
Tell us what gets inspected and who signs off on it. We will tell you what a product like Homesly takes.
Contact MCCSoft