Finding the right leads with AI research

The AI lead-research tool (Find Leads → AI Research) searches the live web for real, contactable people in a market and adds them straight to your CRM. It's tuned for real-estate roles — builders, developers, agents, landlords, property owners, title companies, attorneys — but it isn't limited to them: use the Other (specify) type to find any niche (even outside real estate, e.g. dentists for a marketing list). Quality of results depends entirely on how you frame the search.

1. Pick a tight, real market

The market field should be specific enough to be a real place but broad enough to have a decent supply of contacts.

• Good: "Austin TX", "Denver metro", "Charleston SC", "Phoenix Arizona"
• Too narrow: "78704 zip code", "downtown Austin" — fewer than 10 contacts likely exist publicly online
• Too broad: "Texas", "Midwest" — AI defaults to the biggest cities and ignores the rest

2. Categorize the audience as Buyer, Seller, or Other

This decides which CRM bucket the contacts land in.

• Buyer — people who acquire property (builders, flippers, investors, landlords)
• Seller — people who offload property (property owners, wholesalers selling deals)
• Other — title cos, attorneys, lenders, government — service providers, not deal counterparties

3. Pick contact types thoughtfully

Each checkbox biases the AI toward specific search sources:

• Builders / Developers — biased toward permit databases, construction trade pubs, BuildZoom
• Wholesalers — biased toward off-market real estate forums, MLS sold-listings
• Flippers / Rehabbers — biased toward flip-tracking data + rehab loan filings
• Investors / Funds — biased toward LLC filings, investor directories
• Buy-and-Hold Landlords — biased toward rental ownership records
• Real Estate Agents — biased toward MLS + brokerage directories
• Brokers / Firms — biased toward commercial brokerage rankings
• Property Owners (potential sellers) — biased toward county records, absentee owners
• Property Managers — biased toward PM company directories and rental listings
• Other (specify) — type your own niche (e.g. "probate attorneys", "mortgage note buyers", "land trusts")

Pick 1-3 types per search. Selecting all of them produces a confused mix.

Switching the Categorize as radio to Other swaps the checkbox list to service-provider types instead: Title Companies, Attorneys, Lenders / Hard Money, and City / Government.

4. Use the optional fields

• Property type — only applies to property-relevant types (Builders, Owners, etc.). Hides automatically for Agents/Brokers/PMs/Other.
• Add tag to all found contacts — e.g., "Austin-Q4" so you can filter this batch later in the CRM
• Keywords to mention — e.g., "single-family, infill, distressed" — biases search results toward listings/sources that mention those terms

5. Set max contacts to add

The Max contacts to add field accepts 1–50 (default 10), at roughly 20–30 seconds per contact. A progress bar shows the estimated time left while it runs, and you can cancel anytime.

You'll often get fewer than you asked for — the AI only adds real, verifiable people, so a request for 5 might return 2 if that's all it can confirm in that market. That's expected, not a bug: just run it again to keep accumulating more (each run explores different sources, so you also get more variety). Start with 10–20 on a new market, and don't max out the 50 ceiling on an unproven one — several smaller runs beat one big one.

6. Review, then start outreach

Each found contact comes in with a name/company, email, phone, website, market, and notes on what they do. Right from the results (or the CRM) you can add them to a campaign, log a touch if you've already called, or message them directly. Before you do, sanity-check the new contacts:

• Real name + real company?
• Email looks legit (not info@ or contact@ generics — those are usually fabricated patterns)?
• Market matches?
• Contact type fits?

Tag any duds with "review" and delete after. Direct REI's AI only marks emails it found in real sources (never fabricates), but you'll still see some generics — those are real, but typically lower-conversion.

What's next

With a clean list, head to Writing sequences that get replies.