Put mid-term rental data inside the AI you already use

1. What you plug into the chat

Mid-term rental data is awkward for a general model. The stay is 30 nights or more, usually one to six months. The price is a monthly rent, not a nightly rate with a discount invented on top. The useful comps are other furnished monthly listings, not a 2-night vacation rental on the same street.

MTR Analytics already publishes that dataset: average monthly rate, occupancy, revenue, listing-level comps, Census context, and county employment. The MCP is how an agent reads it on your behalf, then mixes it with the rest of the product. A single question can ask for homes for sale in a price band, keep the ones whose comps clear a monthly number, and underwrite the address you care about, with the county's jobs sitting next to the cash flow.

That is a different product from pasting a market page into a chat. The agent calls the data when the question needs a number, and it can call it again when the question changes city, bedroom count, or price.

2. Underwrite an address

The Property Analyzer is the underwrite for one house: offer price, bedroom-matched furnished monthly rate, long-term rent, and the expense and debt stack. Cash flow, cap rate, cash-on-cash, and DSCR come out of that stack. The MCP runs the same analysis.

Illustrative prompt. The offer price is an example, not a market quote.

Underwrite 1847 E 2700 S, Salt Lake City, UT 84106 at $489,000 as a 3-bedroom furnished monthly rental with 25% down. Tell me the cash flow, the premium over a long-term rent, and whether the county's employment supports a monthly tenant.

Income is the asking rent of comparable 30+ night listings with that bedroom count. Vacancy is an underwriting assumption on top of that rate. The result does not haircut the rate by market occupancy a second time. If you leave the price off, you still get the comp income and an automated value, and the mortgage stays at zero until you name an offer.

The underwrite is saved as an analysis. It does not, by itself, drop the house into your deal pipeline. Saving a deal is still a step you take in the product.

3. Find a purchase that clears a monthly number

Screening a city by hand means a list of homes for sale on one side and a market rent on the other. The connector does that join. You name the place, the price band, and the monthly income you need. It returns active listings and attaches the furnished monthly revenue for that bedroom count.

Illustrative prompt. The prices are examples, not a market quote.

I'm in Salt Lake City. Find active homes between $400,000 and $550,000 that should clear about $3,200 a month on furnished monthly comps. Rank them and underwrite the best 3-bedroom.

The monthly figure on that first pass is the market's comps, not a walkthrough of that kitchen. Screen yield is annual comp revenue divided by the list price. Use it to throw out the weak prices. Then underwrite the address, with your down payment and rate, before you call it a deal.

The same answer can carry the Census picture for the city and BLS employment for the county: renter share, household income, unemployment, and which industries are concentrated there. A travel-nurse or corporate-housing thesis is easier to defend when the jobs are in the same reply as the rent.

4. Check a lease before you sign it

Arbitrage is the other side of a purchase. You lease an unfurnished home and operate it as a furnished monthly rental. The question is whether the furnished revenue covers the asking rent after occupancy, before you spend a weekend on furniture.

Illustrative prompt. The rent is an example, not a market quote.

Find whole-home leases in Austin under $2,200 a month where the furnished monthly revenue still covers the rent after occupancy. Skip rooms in shared houses.

The spread after occupancy is furnished monthly revenue minus the asking rent. The spread before occupancy uses the asking monthly rate with no occupancy haircut. A positive number is a reason to keep reading the lease. It is not cash flow: utilities, furnishing, and the landlord's rules are still yours to check. Rooms are left out, including a cheap bed in a shared house that would otherwise look like a bargain.

5. Compare the deals you already saved

If the houses are already in your account, you should not have to paste them back into a chat. Ask the connector to read the pipeline. Each saved deal comes back with its status, list price, and the Property Analyzer figures when that deal has been underwritten. They are ordered by monthly cash flow. A deal you bookmarked and never ran sits at the end, marked as not underwritten.

Compare the deals in my pipeline. Which ones have the strongest cash flow, which are only saved, and which of the saved ones are in markets with higher occupancy?

That last clause is the mix. The pipeline is yours. Occupancy is the published market. The agent can hold both without you exporting a spreadsheet.

6. Rates, hospitals, Census, and jobs in one answer

The market tools are still there, and they are broader than a single rent figure.

• One place. Average and median monthly rate, occupancy, monthly revenue, listing count, a bedroom breakdown, and the recent monthly trend. Ask by city and state, or by ZIP.
• Comps. Individual 30+ night listings, with rent, occupancy, and a link when one is published.
• A ranking. States, metros, counties, or ZIPs sorted and filtered by rate, occupancy, revenue, supply, income, unemployment, miles to the nearest hospital, or the count of Level I and Level II trauma centers. Those filters combine. A ZIP search can ask for two Level I centers and an average monthly rate around $2,500 (illustrative, not a market quote) in the same question.
• Local drivers on that screen. University, hospital, military base, major employer, and the other saved demand drivers. Ask for ZIPs that have a university and the rent and trauma filters above. This uses places where those drivers have been collected. A ZIP with no saved record is left out. That is not proof it has no campus.
• Hospitals. The facilities in a city, each with its trauma level. The ZIP and metro counts say how many Level I and Level II centers sit inside that boundary.
• Opportunities. When a report has already been saved for the place, the score, demand drivers, and projects. The connector reads that report. It does not start a new one.
• This month's picks. Hot, emerging, undersupplied, cash-flow, and medical-hub metros from the latest release.
• Census. Population, household income, median home value, renter share, and the related ACS fields for that place. Median gross rent in that block is the Census rent, not the furnished rate.
• Employment. County unemployment, wages, and industry concentration from the Bureau of Labor Statistics. A ZIP inherits its county. It is not a ZIP-level jobs report.

What a serious answer separates

Average monthly rate is asking rent. Monthly revenue is that rate times occupancy. Cash flow is an underwrite. Gross yield ranks markets (annual revenue divided by median home value). Screen yield ranks a list price against comp revenue. Mixing those words is how a chat sounds precise and is wrong. The connector labels them so the agent does not have to guess.

Market rates and comps are the latest complete month. A listing that appeared this morning may wait for the next release. For-sale and lease searches are a fresher pull, cached for about a day, and they are still a screen.

7. Corporate housing, travel nurses, acquisitions

The same stay is sold under several names. Corporate housing is a furnished unit for an assignment or a relocation. Travel-nurse housing is a furnished unit for a 13-week contract. Insurance placements and people booking a month at a time are in the same set of listings.

A corporate housing manager comparing Dallas and Phoenix, a host pricing a 2-bedroom in 78704, and an acquirer hunting Salt Lake purchases under a monthly income target are reading one dataset. The MCP is how each of them asks from the tool they already work in, instead of rebuilding the question inside a dashboard every time the city changes.

8. How a Pro member turns it on

Add this server URL in the AI app. The app opens your browser. You sign in to MTR Analytics, click Allow, and the browser returns to the app. Your password stays on our site.

https://mcp.mtranalytics.com/mcp

Cursor, Claude, and ChatGPT each use that URL. The setup page has the config for each client and what to do when Allow does not stick.

The connector is part of Pro. A free account can read the public market pages on this site. Querying the dataset from an AI app, including underwriting and for-sale screens, starts after you upgrade.

Frequently asked questions

What is the MTR Analytics MCP?

It is a remote connector that lets Cursor, Claude, or ChatGPT read MTR Analytics data with your Pro account. The agent can pull furnished monthly rates and comps, Census and county employment, hospital and trauma-center data, a saved Opportunities report, active homes for sale, long-term leases scored against those comps, a Property Analyzer underwrite, and the deals saved in your account.

Can it underwrite a specific house?

Yes. Give the street address and the offer price. The underwrite uses bedroom-matched 30+ night comps for income, a long-term rent for the furnished premium, and the same expense and debt math as the Property Analyzer: cash flow, cap rate, cash-on-cash, and DSCR.

Can it find homes for sale that hit a monthly income target?

Yes. Ask for a city or ZIP, a price range, and the monthly number you want. It returns active listings and estimates furnished monthly revenue from the comps for that bedroom count. Underwrite the address before you treat the screen as a finished deal.

Can it compare deals I already saved?

Yes. Ask it to compare your pipeline. It reads the deals in your account, including cash flow and return figures when a deal has been underwritten, and leaves deals that were only saved at the end of the list.

Does it include Census and BLS data?

When you ask about a place, the answer can include the Census snapshot for that place (income, renter share, home value, and related fields) and Bureau of Labor Statistics employment for the county. Census median rent is not the furnished monthly rate.

Can it use hospital and trauma-center data?

Yes. You can screen metros, counties, or ZIPs by Level I or Level II trauma centers, miles to the nearest hospital, rent, occupancy, income, and the other Market Finder fields in one question. You can also ask for the hospitals in a city and the trauma level of each one. A Level I center that serves a metro may sit in a neighboring city, and the metro count includes it. Adding a university, military base, or major employer uses saved demand drivers. A ZIP with no saved record is left out, which is not proof the place has none.

Can it read Opportunities?

Yes, when a report has already been generated for that place. The agent can read the score, the demand drivers, and the projects, and it can keep one segment such as healthcare. The connector does not start a new Opportunities report.

Which AI apps can connect?

Cursor, Claude, and ChatGPT, plus any client that supports a remote MCP server and OAuth. The setup page has the server URL and the config for each one.

Do I paste an API key?

No, not for the normal path. You add the server URL, the app opens mtranalytics.com, and you click Allow. A key in Settings exists for clients that can only send a header.

Is the connector on the free plan?

No. Public market pages stay free. Querying the dataset from an AI app, including underwriting and for-sale screens, is included with Pro.