The agent storefront at https://agents.homesage.ai/llms.txt?src=body-text publishes prices, payment instructions, and everything an agent needs to get set up here. Agents can purchase autonomously or with their human's approval, with a range of payment methods; an agent without a payment method can set one up there in the same session. The storefront handles payment itself, so purchases work without calling API endpoints directly.
AI underwriting data for private lenders
Send a property address. Our AI reads the listing photos and recent sales nearby. You get the estimated ARV and an estimated renovation budget, plus comps, liens and location risk.
AI underwriting runComplete
Address Listing photosSold recordsLoan records
Vision model reads the listing photos
Condition score of Excellent, Good, Fair or Poor.
AI model
Renovation model prices each room
Line items with a low and high cost, from the photos.
AI model
Comps model ranks similar sales
Estimated ARV with a confidence level.
AI model
Loan records checked
Open loans, estimated equity and loan-to-value.
Records
Underwriting snapshotSample from our docs
Estimated ARV340,000
Estimated renovation cost40,000
Recommended max renovation budget66,200
Estimated LTV on open loans64.77%
Pre-foreclosureNo
Samples from the renovation return and the mortgages and liens docs pages, not the same property. Address hidden.
160M+US properties
30+API endpoints
200B+data points
API and MCPone key for both
Every field below comes back as JSON from one property address. The tag shows whether an AI model or property records produced it.
| Field | What you get | Made by | Endpoint |
|---|---|---|---|
| Estimated ARV | Value after a standard renovation, from similar sales nearby. The search starts at 12 months and 2 miles, then widens to 24 months and 10 miles. | AI model | renovation-return |
| Value before work | Today’s estimated value of the property. | AI model | current-estimate |
| Estimated renovation cost | One total for a standard mid-grade refresh. | AI model | renovation-cost |
| Room-by-room budget | Line items per room with a low and high cost. Major red flags are listed too. | AI model | renovation-cost-breakdown |
| Condition score | Read from the listing photos as Excellent, Good, Fair or Poor. | AI model | property-condition |
| Recommended max renovation budget | 93% of the estimated ARV minus today’s estimated value. | AI model | flip-return |
| Flip projection | Total project cost with estimated profit and estimated ROI for the borrower’s plan. | AI model | flip-return |
| Comparable sales | Sale price, sale date and distance for each comp. Each comp also has a condition and a confidence score. | AI model | comps |
| Open loans and liens | Lender, original amount, estimated balance and origination date for each open loan. | Property records | mortgage-lien-analysis |
| Equity and loan-to-value | Estimated equity, estimated LTV and a pre-foreclosure flag. | Property records | mortgage-lien-analysis |
| Location risk | Nearby risks such as industrial sites and rail lines, with a risk score for the address. | Property records | location-red-flags |
A worked example with the renovation return sample. Swap in your own caps and the borrower’s purchase price.
| Check | Your cap | Sample input | Largest loan |
|---|---|---|---|
| Loan to estimated ARV | 70% | Estimated ARV of 340,000 | 238,000 |
| Loan-to-cost | 90% | Purchase of 250,000 plus estimated renovation of 40,000 | 261,000 |
| Loan you can offer | The lower of the 2 | Both checks above | 238,000 |
You see what each model reads and what it returns, so your team can check the reasoning.
Property addressMatched to 1 of 160M+ US property records.
Listing photosRead by the vision and renovation models.
Sold and loan recordsRecent sales nearby, mortgages and liens.
Market signalsDays on market, price changes and ZIP activity.
Vision condition modelReads the listing photos and scores the condition.
Renovation budget modelAn LLM prices each room from the photos.
Comps modelRanks similar sales and returns an estimated ARV with its confidence.
Return projectionsApplies set cost rules for holding, selling and contingency.
Estimated ARVWith its confidence level.
Renovation budgetTotal and room by room.
CompsEach with a confidence score.
Liens and LTVFrom loan records.
Every value is an estimate for screening and checking a file. When sales are thin, as in non-disclosure states, the response says so with a warning and a low confidence on the estimated ARV.
Run these 4 checks on every loan request. Each one names the endpoints that do it.
1
Start with the borrower’s plan and get the estimated ARV.
Renovation return renovation-return
Value before the work, estimated ARV after it, estimated renovation cost and estimated ROI.
Fix-and-flip return flip-return
Total project cost for the flip, with estimated profit and estimated ROI. It also returns a recommended maximum renovation budget.
2
Compare the borrower’s scope with an independent estimate before you release a draw.
Estimated renovation cost renovation-cost
One estimated renovation cost for a standard mid-grade refresh, with the estimated value increase.
Room-by-room renovation budget renovation-cost-breakdown
A room-by-room renovation estimate from the property photos, with a low and high cost per item and major red flags.
Property condition property-condition
A condition score from the property photos. Scores run Excellent, Good, Fair or Poor.
3
Check the estimated ARV against what similar properties sold for nearby.
Comparable sales comps
Recent sold properties near the subject with sale price, sale date and distance. It also returns an estimated ARV.
Current estimate current-estimate
Today’s estimated value for the property, as one number.
4
Find open loans and nearby risks before you take a lien position.
Mortgages and liens mortgage-lien-analysis
Open loans with lender, estimated balance and start date. It also returns estimated equity and a pre-foreclosure flag.
Location red flags location-red-flags
Nearby risks such as industrial sites, rail lines and power lines. It also returns a risk score for the address.
The renovation return endpoint gives the value before work, the estimated ARV, the estimated renovation cost and the estimated ROI.
Add comps and liens when the file needs them.
cURLPythonJavaScript
curl -G "https://api.homesage.ai/api/properties/renovation_return/" \ -H "Authorization: Bearer $HOMESAGE_API_KEY" \ --data-urlencode "property_address="
import os, requests
r = requests.get(
"https://api.homesage.ai/api/properties/renovation_return/",
headers={"Authorization": f"Bearer {os.environ['HOMESAGE_API_KEY']}"},
params={"property_address": ""},
)
print(r.json()["post_renovation_arv"])
const url = new URL("https://api.homesage.ai/api/properties/renovation_return/");
url.searchParams.set("property_address", "");
const res = await fetch(url, {
headers: { Authorization: `Bearer ${process.env.HOMESAGE_API_KEY}` },
});
const { post_renovation_arv } = await res.json();
Response
{
"pre_renovation_avm": 250000,
"post_renovation_arv": 340000,
"renovation_cost": 40000,
"net_return": 50000,
"roi_percent": 125,
"max_recommended_renovation_budget": 66200
}
Sample response from the renovation return docs page, trimmed. Estimated values.
| Option | Best for | Start here |
|---|---|---|
| REST API | Your own software, models and data pipelines | API docs, one page per endpoint |
| MCP server | Loan officers who work in Claude, ChatGPT or another AI assistant | Connect the MCP in 3 steps |
| Full Property Report | One file checked by hand, no code | See the Full Property Report |
| Sage | Questions in plain English inside the Homesage.ai app | Meet Sage |
The core checks are the after-repair value, the rehab budget, comparable sales and existing liens. Homesage.ai returns data for each from one address.
It comes from recent sales of similar properties nearby. The search starts at 12 months and 2 miles and widens to 24 months and 10 miles when sales are thin.
The mortgages and liens endpoint returns estimated equity and an estimated LTV for open loans. To see your loan as a share of the estimated ARV, divide the loan amount by it.
The values are model estimates for screening and checking files. Order an appraisal wherever your credit policy needs one.
The vision and renovation models read the photos on the property’s listing.
Values are estimates from models and do not replace an appraisal.
The models only see what the listing photos show.
Coverage is the United States only.
Thin sales data lowers the confidence on the estimated ARV, and the response tells you.
Usage is credit-based. Plans are on the pricing page.
One set of APIs for records, rents, comps and AI fields such as condition from photos.
Screen a city, county or ZIP against your buy box, with AI scores on every match.
AI-ranked comps, estimated values and ZIP market outlook for every listing.
How lenders use our web products alongside the APIs.
One report per property, for files your team checks by hand.
7 best APIs for hard money lenders
How the main property data APIs compare for flip loans.
How the estimated ARV is worked out.
How to check if a property has a lien
The steps behind the mortgages and liens check.
Create an account and test the endpoints on a property you are reviewing this week. Or book a demo with our team.
Estimated values. Not investment advice. Sample responses come from our public API docs, read 2026-11-07.