Property Research AI — Paul Slavinski

The Problem

You spend hours each day manually cross-referencing three databases to research properties — pulling ownership history from the Registry of Deeds, assessed values from Cambridge's database, and comps from MLS. That's skilled work trapped in a data-entry loop.

What if you typed an address and got a formatted property brief in seconds?

Your Three Databases — Automation Feasibility

Database	Best For	Feasibility
Cambridge Assessors Open Data Portal (Socrata API)	Assessed values, property details, owner info, sale history — 62 data fields going back to FY2016	Easy Free public API, clean data
MLS PIN RESO Web API / IDX Feeds	Comps, listing history, market context — up to 5 years of sold data	Medium Official API available via broker agreement
Middlesex Registry of Deeds MassLandRecords	Official deeds, mortgages, liens, ownership transfers, tenure calculation	Hard CAPTCHA-protected, no public API

What We Found

Quick Win

🏛️ Cambridge Assessors — Wide Open

Cambridge publishes their entire property database on an Open Data Portal with a free API. No scraping needed — structured queries returning clean JSON. We can pull 62 fields per property instantly:

owner_name assessedvalue saleprice saledate book_page interior_livingarea interior_bedrooms interior_fullbaths condition_yearbuilt landarea zoning latitude/longitude

Historical data available FY2016–FY2023. This is the fastest win — we can have this working in days.

Official Integration

📊 MLS PIN — Structured Access Available

MLS PIN is RESO Web API certified (the modern standard for MLS data). They offer multiple official paths:

Free IDX — sold search up to 5 years back, pipe-delimited downloads + FTP photos
Brokerage access — $100/month, richer data for internal analytics
RESO Web API — standards-based programmatic access

If Senné already has an MLS PIN agreement, we may be able to piggyback on existing access for the integration.

Assisted Workflow

📜 Registry of Deeds — Smart Assist, Not Full Automation

MassLandRecords is protected by Imperva + hCaptcha — no public API, active bot prevention. Full automation isn't practical, but we can still save significant time:

Pre-fill searches using owner name + book/page from the assessor data
Auto-extract deed metadata once you're past the CAPTCHA
Calculate ownership tenure from recording dates
Flag edge cases — trusts, LLCs, intra-family transfers

You clear one CAPTCHA, the AI handles the rest of that session.

The Proposed Workflow

Type an address. Get a property brief. Here's what happens behind the scenes:

Step 1

Address → Parcel Identification

Normalize the address, resolve to Cambridge parcel ID via Open Data API

Step 2

Pull Assessor Snapshot

Assessed value, owner name, property details, last sale date/price, year built, condition — all 62 fields

Step 3

Registry Lookup + Tenure

Use book/page from assessor data to locate the deed. Calculate how long the current owner has held the property. Flag trusts or unusual ownership.

Step 4

Pull Comps

Nearby sold properties matched by class, size, beds/baths. Compute $/sqft, medians, and adjustments.

Step 5

Generate Property Brief

One-page PDF with executive summary, assessor details, ownership timeline, comp table, and red flags.

Build Timeline

Phase 0

2–4 days

Clickable prototype + Cambridge assessor pull

Phase 1

1–2 weeks

Robust Cambridge automation + report template

Phase 2

2–4 weeks

MLS PIN integration via official feeds

Phase 3

2–4 weeks

Registry assist + tenure calculation

Questions for Paul

Does Senné already have MLS PIN data access (IDX, VOW, or RESO Web API)?
Do you mostly work with Recorded Land, or do you also need Land Court records?
Preferred output: PDF one-pager, email, or directly into a CRM?
How deep should ownership history go — last deed only, or full 10+ year chain of title?
Do you research properties outside Cambridge (Somerville, Boston, etc.)?

Ready to Stop Tab-Juggling?

We can have a working prototype pulling Cambridge assessor data within days. You focus on strategy — let the AI handle the data grunt work.

[email protected] · feralbots.com