Which NYC Mortgage Brokers Does AI Recommend? 2026

Study date: November 12, 2026

When we asked ChatGPT, Perplexity and Google AI Overview 10 everyday questions about getting a mortgage in New York City, GuardHill Financial was the name they gave most, with 16 mentions. It was also named first in an answer more than any other business, 5 times. R&J Capital Group (13), Chase (11), Block Financial Resources (10) and Grayton Mortgage (8) came next.

In all, the 28 answers named 15 mortgage brokers and lenders 92 times. That's a very different picture from our NYC lawyers study the same day, where the engines named almost no firms. For mortgages, they named businesses readily, but the source they cited most was the state regulator.

These are the answers the engines returned in one Beket AI audit on one day. It isn't our ranking of any broker or lender, it isn't financial advice, and AI answers change over time.

Key numbers

Measure	Result
Study date	November 12, 2026
Location setting	New York City, New York, United States
Questions	10 unbranded mortgage questions, in English
Engines	ChatGPT, Perplexity, Google AI Overview
Answers returned	28 of 30 (ChatGPT 10, Perplexity 10, Google AI Overview 8)
Brokers and lenders named	15
Mentions	92
Named on all three engines	4
Citation events	268
Most-cited domain	dfs.ny.gov, 21 citation events

The names AI gave most

Rank	Broker or lender	Mentions	Times named first	ChatGPT	Perplexity	Google AI Overview
1	GuardHill Financial	16	5	11	3	2
2	R&J Capital Group	13	0	8	4	1
3	Chase	11	1	3	6	2
4	Block Financial Resources	10	1	8	0	2
5	Grayton Mortgage	8	0	8	0	0
	10 other brokers and lenders	34	5
	All 15	92	12	54	24	14

Source: Beket AI market audit, November 12, 2026. Four R&J Capital name variants were combined into R&J Capital Group.

The answers mixed independent brokers with banks. Chase, Citibank and Wells Fargo all showed up, and so did National Cooperative Bank, which came up on the co-op questions. We kept banks and lenders in the count because several questions asked for lenders, and the engines offered them as options.

Four names were mentioned on all three engines: GuardHill Financial, R&J Capital Group, Chase and National Cooperative Bank. Seven more were named on two engines, and four on only one.

Engine by engine

Engine	Questions answered	Citation events	Names	Mentions	Name given most
ChatGPT	10/10	48	12	54	GuardHill Financial (11)
Perplexity	10/10	160	10	24	Chase (6)
Google AI Overview	8/10	60	8	14	Six names tied at 2

ChatGPT did most of the naming, with 54 of the 92 mentions, while citing the fewest sources. Its top four (GuardHill Financial 11, then Block Financial Resources, Grayton Mortgage and R&J Capital Group with 8 each) were all independent brokers. Grayton Mortgage and The Funding Store were named only by ChatGPT.

Perplexity cited the most sources, 160 across 10 answers, but named businesses less often. Its top name was a bank, Chase, with 6 mentions, followed by R&J Capital Group (4), Alpine Mortgage (3) and GuardHill Financial (3).

Google AI Overview answered 8 of the 10 questions and spread its 14 mentions thinly: Block Financial Resources, Chase, Citibank, GuardHill Financial, MortgageDepot and Wells Fargo had 2 each. It returned no answer for two questions, mortgage brokers for self-employed borrowers and FHA loan lenders. We list those as missing, not as zero mentions, and the tools don't show why they're missing.

The run started shortly after Google said it was moving Search to its Gemini 4 Hydrogen model on November 12. The audit tools don't report which model wrote each answer, so treat the Google AI Overview numbers here as a snapshot from that day, not a confirmed result for one model.

How New York compares

Next to our earlier mortgage studies, New York's answers named a little fewer businesses per answer, and the leader held a bigger share.

Study	Date	Questions	Answers	Businesses named	Mentions	Mentions per answer	Leader's share
NYC mortgage brokers	Nov 12, 2026	10	28	15	92	3.3	17.4%
Seattle mortgage brokers	Oct 27, 2026	25 + 4 discovery	76	39	311	4.1	12.9%
Vancouver mortgage brokers	Oct 23, 2026	25 + 4 discovery	63	35	236	3.7	10.6%

These studies used different question sets, sizes and dates, so read this as a rough comparison, not a controlled test. One pattern held in all three: ChatGPT gave the most mentions, and Google AI Overview answered the fewest questions.

Where the answers pointed

The top cited domain was dfs.ny.gov, the New York State Department of Financial Services, which supervises the state's mortgage bankers and brokers, with 21 citation events. Add the federal Consumer Financial Protection Bureau (consumerfinance.gov, 9) and the NMLS Consumer Access licensing lookup (nmlsconsumeraccess.org, 7), and regulator and government sources had 37 citation events in the top ten.

Lender and broker websites matched that, with 37: Chase's home lending pages had 18, guardhill.com 10 and rjcmortgage.com 9. The two brokers whose own sites were cited most were also the top two independent names in the answers. Media and ranking sites (Brick Underground, Expertise.com and Bankrate) had 22.

Domain	Citation events
dfs.ny.gov	21
homelendingadvisor.chase.com	18
google.com	17
guardhill.com	10
consumerfinance.gov	9
rjcmortgage.com	9
brickunderground.com	8
expertise.com	8
nmlsconsumeraccess.org	7
bankrate.com	6

The top ten domains cover 113 of the 268 citation events, so this is the head of the list, not all of it.

Why this matters for NYC brokers

Buyers already use AI for this. In a Veterans United survey released in February 2025, 32% of prospective homebuyers said they use AI tools in their homebuying journey, and 28% of those use them to shop for the best mortgage rates.

The buyers are also harder to win. The NAR 2025 Profile of Home Buyers and Sellers found first-time buyers fell to a record-low 21% of all buyers, with a record-high median age of 40. One of our 10 questions was written for first-time buyers, and two were about buying a co-op, a New York staple.

In this run, the brokers that came up most also had their own pages cited, and the engines leaned on the regulator for the rules. That suggests two things to check: whether your own pages answer buyer questions plainly, and whether your public license record is easy to find and accurate.

What an NYC broker can check

1. Ask unbranded questions for your niche. Co-op loans, jumbo loans and self-employed borrowers each got different names here. Our guide to checking whether ChatGPT recommends your business shows how to do it by hand.
2. Check each engine on its own. ChatGPT leaned toward independent brokers, Perplexity named a bank first, and Google AI Overview spread its mentions thinly.
3. Look at what gets cited. If answers cite dfs.ny.gov or NMLS Consumer Access, make sure your license details there match your website.
4. Write the page AI can quote. One cited GuardHill page covered complex mortgages for high-net-worth clients. A clear page on, say, "how a co-op loan works in New York City" gives an engine something specific to use.
5. Re-check over time. One run is a snapshot.

The free Beket AI audit runs 25 prompts across ChatGPT, Perplexity and Google AI Overview for your own brand. Beket AI Pro is $49 a month with daily re-audits, starting at 50 prompts across the same three engines.

Method and caveats

Beket AI ran one market audit for mortgage brokers in New York City on November 12, 2026, created at 12:36 PM PT and completed at 12:43 PM PT. We used 10 unbranded questions in English, set the location to New York City, New York, United States, and used three engines: ChatGPT, Perplexity and Google AI Overview. Extra prompt generation was off, so 30 answers were requested. We picked the 10 questions for real buyer intent: the main loan types and buyer situations in New York, plus cost and how to choose.

The final answer log has 28 answers, with no duplicate question and engine pairs and no off-topic rows. ChatGPT and Perplexity answered all 10 questions. Google AI Overview returned answers for 8.

The raw market report had 20 labels and 98 mentions. We left out Zillow (3 mentions, all Google AI Overview), a listing and lender-directory site, and Fannie Mae (3 mentions), which works in the secondary mortgage market and doesn't lend to consumers directly. We merged four R&J Capital name variants into R&J Capital Group, the name on its website. That left 15 brokers and lenders with 92 mentions. These are editorial calls. NerdWallet and Bankrate weren't extracted as names, so we report no mention counts for them.

Counts are mention events, not recommendation rates. One answer can name a business more than once. Being named doesn't verify a New York license, an NYC office or the quality of anyone's loans. The citation domain counts come from the market report's top ten, and the citation URLs we inspected were a sample of 100 of 268. We didn't add names from citation URLs that the report didn't extract.

AI answers change over time, and this is a single run. If your business is in this study, or you'd like it re-run, you can ask Beket AI for another run.

All 10 questions

#	Question	Google AI Overview answered?
1	best mortgage brokers in New York City	Yes
2	mortgage brokers for first-time home buyers in New York City	Yes
3	co-op loan lenders in New York City	Yes
4	mortgage broker vs bank for buying a co-op in New York City	Yes
5	jumbo mortgage brokers in New York City	Yes
6	mortgage brokers for self-employed borrowers in New York City	No answer
7	FHA loan lenders in New York City	No answer
8	refinance mortgage brokers in New York City	Yes
9	how much does a mortgage broker cost in New York	Yes
10	how to choose a mortgage broker in New York City	Yes

ChatGPT and Perplexity answered all 10.

Bottom line

For New York City mortgage questions, the engines in this run did name businesses, and a handful of independent brokers led, with GuardHill Financial and R&J Capital Group the only brokers named on all three engines. Each engine still had its own list, and the source they trusted most was the state regulator. For an NYC broker, that means checking each engine with the questions buyers actually ask, and making sure both your own pages and your public license record tell the same clear story.