ProgrammableMarketing/chestnutmortgage.com (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 5
  • This quarter · 11
  • Big bets · 3
  • Wildcards · 2
Generated Nov 10, 2026
  1. Programmable
  2. Playbooks
  3. Chestnut (Detta, Inc. dba Chestnut Mortgage)
Chestnut (Detta, Inc. dba Chestnut Mortgage) logo

growth playbook · Nov 10, 2026

chestnutmortgage.com (opens in a new tab)

Chestnut (Detta, Inc. dba Chestnut Mortgage) growth playbook

Diagnosis

Chestnut is a two-person Y Combinator (Spring 2025) company with a sharp promise: AI agents that save borrowers 0.5%+ in interest, licensed in Texas and Colorado and hiring its first engineer to extend the voice agent and improve quote close rates.

The biggest problem is upstream of all of that: chestnutmortgage.com answers Google's own tester with a 403 error page, no pages from the site are in search, the name "Chestnut Mortgage" leads to a UK broker's reviews, and a directory already shows a placeholder address.

This playbook first makes the site findable and the brand unmistakable, then makes the quote convert (public sample rates, a proven savings claim, five-minute voice callbacks, rate alerts), builds Texas and Colorado reach through agents, metro pages and reviews, and uses the 100+-lender data for bigger bets: a next-state decision from demand, a visible no-lender-fee price and a public rate-spread report.

Growth Score

6/10

Biggest Opportunity

A rate-shopping agent across 100+ lenders produces data no single lender has: put it in public (sample rates, a second opinion on any Loan Estimate, a weekly spread report) so that every borrower in Texas and Colorado who shops has a reason to start with Chestnut.

Biggest Risk

A site that search engines cannot read and a name shared with an established UK broker mean that early word of mouth leaks away, while Own Up, Tomo and Better already make similar savings, speed and AI promises with thousands of reviews behind them.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Let Google in: chestnutmortgage.com answers Google's own tester with a 403 error page, and none of its pages are in search

    Run Search Console URL Inspection on the homepage and record the HTTP status Google receives.

    High impact First result in 1-2 weeks
  2. 2

    Own the name "Chestnut Mortgage" in search before a UK broker's reviews define it

    Write the one-line company description and the exact name to use everywhere.

    High impact First result in 1 month
  3. 3

    Put the "0.5%+ saved" claim on a methodology page with worked examples

    Write down the benchmark the 0.5% figure is measured against.

    Medium impact First result in 1 month

02

Your idea portfolio

21 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 7 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 3Big bets · 8Fill-ins · 6Reconsider · 4
Effort →

Showing 21 of 21 ideas

This week 5

Quick wins with a clear first action.

  • 01

    Chestnut's whole funnel ends at "get an instant quote on the website", but on November 10, 2026 Google's PageSpeed tester got HTTP 403 and a Cloudflare error page instead of the homepage, and flagged "Page is blocked from indexing". The homepage, the www address, the http address and robots.txt all returned 403, and a site: search finds no chestnutmortgage.com pages at all, only directories and Y Combinator pages. Someone who hears about Chestnut and searches for it cannot reach the quote page from search, and third-party directories fill the gap, sometimes wrongly.

    High impactLow effort First result in 1-2 weeks Software Engineer

    First action

    Run Search Console URL Inspection on the homepage and record the HTTP status Google receives.

    How to measure it

    Indexed pages and brand clicks

    Now
    –
    Target
    Homepage and quote page indexed; 50 brand-search clicks a month
    By
    30 days

    Steps

    1. 1

      Reproduce

      Load the homepage from a clean browser, a phone on mobile data and PageSpeed; note who gets the 403.

      Founder/CEO 1 hour

    2. 2

      Find the rule

      Check the CDN's security events for blocked requests to / and /robots.txt; review bot, country and challenge rules and the root-domain DNS setup.

      Software Engineer 2 hours

    3. 3

      Allow verified crawlers

      Let verified search bots and Google's testing tools reach public pages; keep rate limits and challenges on quote and application endpoints.

      Software Engineer 2 hours

    4. 4

      Publish robots.txt and sitemap

      A robots.txt that returns 200 and points to sitemap.xml listing the homepage, quote, state, licensing and product pages.

      Software Engineer 2 hours

    5. 5

      Add the basics

      Title, meta description, share image, favicon and Organization markup with the legal name, dba, states and products.

      Founder/CEO 1 day

    6. 6

      Verify and submit

      Search Console and Bing Webmaster Tools: verify, submit the sitemap, inspect and request indexing.

      Software Engineer 1 day

    7. 7

      Re-test

      Re-run PageSpeed until it shows the real homepage; record speed and accessibility.

      Software Engineer 1 hour

    8. 8

      Monitor

      An outside uptime check on the homepage and quote page that alerts on any 403 or 5xx.

      Software Engineer 2 hours, then weekly

    First result
    1-2 weeks
    Cost
    $0 plus a day of engineering time
    Owner
    Software Engineer

    Smallest test

    Allow verified crawlers on the homepage only today, re-run PageSpeed and run URL Inspection.

    Why it works

    Find the rule that turns automated visitors away (a bot or firewall setting, a country rule, or a DNS/proxy conflict on the root domain) and let verified search crawlers and Google's testing tools through, while keeping protection on the quote and application flows. Then confirm with Search Console URL Inspection, publish a robots.txt and a sitemap (homepage, quote, Texas, Colorado, licensing and product pages), and add a title, meta description, share image and Organization markup naming Detta, Inc. dba Chestnut Mortgage, the two states and the products. Re-run PageSpeed until it shows the real homepage, which also gives the first real-visitor speed data.

    For a two-person company, brand search and AI-assistant answers are the cheapest acquisition there is: every person who sees the launch post, the demo video or a friend's recommendation checks a search engine first. Today that path ends at a UK broker or a directory with a placeholder address. Fixing access takes hours and unblocks every content idea in this playbook.

    Ready to use

    ChecklistSearch access checklist

    [ ] Homepage returns 200 to PageSpeed and Search Console URL Inspection
    [ ] /robots.txt returns 200 and lists the sitemap
    [ ] sitemap.xml: homepage, quote, Texas, Colorado, licensing, products
    [ ] Title and meta description name Chestnut, Texas and Colorado
    [ ] Organization markup: Detta, Inc. dba Chestnut Mortgage, states, products, profiles
    [ ] Share image and favicon
    [ ] Quote and application endpoints still protected
    [ ] Outside uptime alert on 403/5xx

    Risks and how to handle them

    • Scrapers hit the rate quotes once bots are allowed

      Allow only verified crawlers on public pages and keep challenges and rate limits on quote and application endpoints.

    • The block was deliberate while the site is unfinished

      Publish at least a holding page with the states, licensing and a contact route, so the brand name resolves to Chestnut.

    Evidence

    • Your websiteFetches of https://chestnutmortgage.com/, https://www.chestnutmortgage.com/, http://chestnutmortgage.com/ and /robots.txt all returned HTTP 403 on November 10, 2026; a site: search for chestnutmortgage.com returned no pages from the site itself, only third-party directories and Y Combinator pages.Source
    • PageSpeedGoogle PageSpeed test of chestnutmortgage.com (November 10, 2026): the tester received HTTP 403 and a Cloudflare error page instead of the homepage; SEO score 40, flagging "Page is blocked from indexing", "Page has unsuccessful HTTP status code" and a missing meta description; no real-visitor (field) data is available.Source
    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
  • 02

    A search for "Chestnut Mortgage" reviews returns mostly Trustpilot pages for Chestnut Mortgages, a UK broker with 4.8 from 46 reviews, and no US profile for Chestnut. The legal entity is Detta, Inc. dba Chestnut Mortgage, while YC lists the company as plain "Chestnut", so the name is split three ways.

    High impactLow effort First result in 1 month Founder/CEO

    First action

    Write the one-line company description and the exact name to use everywhere.

    How to measure it

    Owned brand results

    Now
    –
    Target
    6 of the top 10 results for "Chestnut Mortgage" in the US
    By
    60 days

    Steps

    1. 1

      Set the name

      "Chestnut Mortgage" plus a one-line description with Texas, Colorado and the products.

      Founder/CEO 1 hour

    2. 2

      Business profile

      Create or claim the Google Business Profile for the San Francisco office.

      Founder/CEO 1 day

    3. 3

      Align profiles

      LinkedIn, Crunchbase, YC, X and Facebook use the same name, description and link.

      Founder/CEO 2 hours

    4. 4

      Add sameAs markup

      Organization markup on the site linking every profile.

      Software Engineer 1 hour

    5. 5

      Open a review profile

      Google first, then Zillow, linked from the closing email.

      Founder/CEO 1 day

    6. 6

      Track

      Monthly check of the top 10 US results for the brand name.

      Founder/CEO monthly

    First result
    1 month
    Cost
    $0
    Owner
    Founder/CEO

    Smallest test

    Update the YC and LinkedIn descriptions today and recheck results in two weeks.

    Why it works

    Pick one public name ("Chestnut Mortgage") and use it everywhere: site title, Google Business Profile for the San Francisco office, LinkedIn, Crunchbase, YC profile, the email footer and the licensing disclosure. Link them with sameAs markup. Open a US review profile (Google, then Zillow) so the first star rating searchers see is Chestnut's own. Add "Texas and Colorado" to every profile description to separate it from the UK firm.

    Borrowers check reviews before sharing income and credit details. A consistent name and an owned profile stop the UK firm's page from answering that check.

    Risks and how to handle them

    • Search engines keep mixing the two firms

      Always pair the name with "Texas and Colorado" and the San Francisco address.

    Builds on

    Evidence

    • Search resultsSearching "Chestnut Mortgage" reviews returns mostly Trustpilot pages for Chestnut Mortgages, a UK mortgage broker at chestnutmortgages.co.uk (4.8 from 46 reviews); no US review profile for Chestnut appeared.Source
    • Public recordsSAM.gov registration: DETTA, INC. (dba CHESTNUT MORTGAGE), San Francisco, a Delaware corporation, active since May 6, 2026, registered for federal assistance awards.Source
    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
  • 03

    With the site unreadable, directories write their own version of Chestnut. Extruct's profile is titled "Chestnut Funding", gives the address as "1234 Example Street, Los Angeles" and lists two directors whose names appear in the UK broker's reviews, next to accurate details such as quotes in under 2 minutes and 100+ lenders.

    Medium impactLow effort First result in 2-4 weeks Founder/CEO

    First action

    Send Extruct a correction with the right name, address and founder.

    How to measure it

    Correct profiles

    Now
    –
    Target
    All known profiles correct
    By
    30 days

    Steps

    1. 1

      List profiles

      Search the domain and brand name; record each profile and what's wrong.

      Founder/CEO 2 hours

    2. 2

      Write the fact sheet

      Name, legal entity, address, states, products, founder, licensing.

      Founder/CEO 2 hours

    3. 3

      Publish it

      An About/Press page on the site.

      Software Engineer 2 hours

    4. 4

      Request corrections

      Use each provider's claim or correction form.

      Founder/CEO 1 day

    5. 5

      Claim where possible

      Claim profiles so future edits come from Chestnut.

      Founder/CEO 1 day

    6. 6

      Recheck

      Quarterly sweep for new profiles.

      Founder/CEO quarterly

    First result
    2-4 weeks
    Cost
    $0
    Owner
    Founder/CEO

    Smallest test

    Fix the Extruct profile this week.

    Why it works

    List every directory and data provider that profiles chestnutmortgage.com (Extruct, Crunchbase, startup databases, mortgage directories), and send each the correct name, address, states, products and founder. Keep a one-page fact sheet on the site (an About or Press page) as the source they can copy from.

    AI assistants and comparison sites copy these profiles. One wrong address or the wrong people on a lender's profile is a trust problem in a category where borrowers check legitimacy.

    Risks and how to handle them

    • Providers ignore requests

      Claim the profile where possible and keep the on-site fact sheet as the source of truth.

    Evidence

    • Directory listingExtruct's profile of chestnutmortgage.com is titled "Chestnut Funding", lists "Instant quoting in under 2 minutes", 100+ lenders, closings in 2 to 3 weeks and typical savings of about 0.5%, but gives the address as "1234 Example Street, Los Angeles" and lists two directors whose names appear in the UK broker's reviews.Source
    • Your websiteFetches of https://chestnutmortgage.com/, https://www.chestnutmortgage.com/, http://chestnutmortgage.com/ and /robots.txt all returned HTTP 403 on November 10, 2026; a site: search for chestnutmortgage.com returned no pages from the site itself, only third-party directories and Y Combinator pages.Source
  • 04

    Tomo publishes its rates, with state rate pages that include Texas and Colorado. Chestnut's rate-shopping agent already sees offers from 100+ lenders, and the new engineer's brief is to "improve close rates with better rate quotes", but a borrower has to start a quote to see a number.

    High impactMedium effort First result in 1 month Founder/CEO

    First action

    Define the example borrower and pull today's offers for it.

    How to measure it

    Visitor-to-quote conversion

    Now
    –
    Target
    +30% on visitors who see the rate table
    By
    60 days

    Steps

    1. 1

      Define the example

      Loan amount, down payment, credit band and property type per product.

      Founder/CEO 1 day

    2. 2

      Pull the data

      Best and median offer from the rate-shopping agent, daily.

      Software Engineer 3 days

    3. 3

      Compliance review

      APR, assumptions and disclosure wording checked by compliance counsel.

      Compliance counsel 3 days

    4. 4

      Build the pages

      /rates/texas and /rates/colorado with a quote button.

      Software Engineer 2 days

    5. 5

      Add to sitemap

      Index both pages.

      Software Engineer 1 hour

    6. 6

      Measure

      Quote starts from the rate pages against other entry points.

      Founder/CEO weekly

    First result
    1 month
    Cost
    Engineering time and a compliance review
    Owner
    Founder/CEO

    Smallest test

    Publish one 30-year fixed Texas sample rate, updated daily, for two weeks.

    Why it works

    Publish a daily sample-rate table for Texas and Colorado (30-year, 15-year, cash-out refinance, HELOC) for a stated example borrower, with APR, points and the date, and a button to get a personal quote. Show the best rate Chestnut found and the median across lenders it checked, so the savings are visible before sign-up.

    Rate shoppers compare published numbers first. Showing them, with the gap against the median, turns the 0.5% claim into something a visitor can see in five seconds.

    Risks and how to handle them

    • Sample rates don't match the visitor's own quote

      State the assumptions clearly and link straight to a personal quote.

    Builds on

    Evidence

    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • HiringSoftware Engineer (San Francisco, in person; $100K-$200K plus 0.5%-3% equity; current team of 2): "Extend the AI voice agent to qualify more leads", "Improve close rates with better rate quotes" and "Automate loan processing with browser agents and automated underwriting"; founded January 2025, backed by Y Combinator and Olive Tree Capital.Source
  • 05

    Chestnut's headline claim is that its agents "save customers 0.5%+ in interest". Tomo makes nearly the same claim (about 0.5% below average), and Own Up backs its "average of $30,000" savings with a linked methodology. Without one, Chestnut's number reads the same as everyone else's.

    Medium impactLow effort First result in 1 month Founder/CEO

    First action

    Write down the benchmark the 0.5% figure is measured against.

    How to measure it

    Methodology-to-quote rate

    Now
    –
    Target
    15% of methodology visitors start a quote
    By
    60 days

    Steps

    1. 1

      Fix the benchmark

      Pick the comparison: a public rate survey or the median offer checked.

      Founder/CEO 1 day

    2. 2

      Pull examples

      Two or three anonymised quotes with best and median offers.

      Software Engineer 1 day

    3. 3

      Write the page

      Method, dollars on a sample loan, limits of the claim.

      Founder/CEO 1 day

    4. 4

      Compliance review

      Claim wording and disclosures.

      Compliance counsel 3 days

    5. 5

      Link it

      From the homepage claim, quote flow and profiles.

      Software Engineer 2 hours

    6. 6

      Refresh

      Update the figures quarterly.

      Founder/CEO quarterly

    First result
    1 month
    Cost
    Writing time and a compliance review
    Owner
    Founder/CEO

    Smallest test

    Add one worked example under the claim on the homepage.

    Why it works

    Publish a short page that says what 0.5% is measured against (a named rate survey or the median offer across the lenders Chestnut checks), how it was calculated, over which loans, and what it means in dollars on a $400,000 loan. Add two or three anonymised worked examples from real quotes. Link the page wherever the claim appears.

    Borrowers and agents are sceptical of savings claims; a clear method turns a slogan into a reason to try the quote.

    Risks and how to handle them

    • The measured saving is lower than 0.5%

      Publish the real figure; an honest smaller number beats an unproven bigger one.

    Evidence

    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
    • Competitor researchOwn Up: "Shop for your mortgage and save thousands"; customers "could save an average of $30,000 over the life of their loan", with a linked methodology; 4.9 from 4,690 verified reviews; free, "No credit impact"; a section for Realtors; says it is not a lender.Source
    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source

This quarter 11

The core plays for the next three months.

  • 06

    Chestnut can only lend in Texas and Colorado, so every visitor from another state is wasted, and local intent ("Austin refinance rates", "Denver mortgage lender") is the most valuable search. Tomo already has state rate pages for both states, and no Chestnut pages are indexed yet.

    High impactMedium effort First result in 2-3 months Founder/CEO

    First action

    Pick the first three metros by quote volume.

    How to measure it

    Metro-page quote starts

    Now
    –
    Target
    100 a month
    By
    1 quarter

    Steps

    1. 1

      Pick metros

      Three to start, by quotes and population.

      Founder/CEO 1 day

    2. 2

      Template

      Sample rate, local facts, products, quote button.

      Software Engineer 1 week

    3. 3

      Write local content

      Questions buyers in each metro ask, answered plainly.

      Founder/CEO 2 weeks

    4. 4

      Compliance check

      Licensing and rate disclosures on each page.

      Compliance counsel 1 week

    5. 5

      Publish

      Add to sitemap and link from the state rate pages.

      Software Engineer 1 day

    6. 6

      Expand

      Add metros where the first three rank.

      Founder/CEO monthly

    First result
    2-3 months
    Cost
    Writing and engineering time
    Owner
    Founder/CEO

    Smallest test

    One Austin page for six weeks.

    Why it works

    Build pages for the main metros in each state (for example Houston, Dallas-Fort Worth, Austin, San Antonio, Denver, Colorado Springs): the local sample rate from the rate table, typical loan sizes, the local property-tax and HOA questions buyers ask, HELOC and cash-out options, and the quote button. Start with three and expand on what ranks.

    A lender licensed in two states should win local searches in those states; metro pages match how buyers search and filter out visitors Chestnut can't serve.

    Risks and how to handle them

    • Thin, repetitive pages

      Only publish a metro when it has its own data and questions.

    Builds on

    Evidence

    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
    • Your websiteFetches of https://chestnutmortgage.com/, https://www.chestnutmortgage.com/, http://chestnutmortgage.com/ and /robots.txt all returned HTTP 403 on November 10, 2026; a site: search for chestnutmortgage.com returned no pages from the site itself, only third-party directories and Y Combinator pages.Source
  • 07

    Rate-drop alerts are one of Chestnut's four launch features, but most mortgage visitors are months from applying. Asking them for a full quote loses them; asking for an email to watch a target rate keeps them.

    Medium impactLow effort First result in 1-2 months Software Engineer

    First action

    Add a target-rate field to the existing alert feature.

    How to measure it

    Alerts that become quotes

    Now
    –
    Target
    500 alert sign-ups, 10% to quote
    By
    1 quarter

    Steps

    1. 1

      Design the form

      Contact, state, loan type, target rate.

      Software Engineer 2 days

    2. 2

      Place it

      Rate, metro and quote-exit pages.

      Software Engineer 1 day

    3. 3

      Write the alert

      Rate hit, what it means in dollars, quote link.

      Founder/CEO 1 day

    4. 4

      Consent and compliance

      Opt-in wording for email and texts reviewed.

      Compliance counsel 3 days

    5. 5

      Connect the voice agent

      Replies trigger a callback.

      Software Engineer 1 week

    6. 6

      Measure

      Sign-ups, alerts sent, quotes started.

      Founder/CEO monthly

    First result
    1-2 months
    Cost
    Engineering time and email/SMS costs
    Owner
    Software Engineer

    Smallest test

    Add the alert form to the quote exit for two weeks.

    Why it works

    Offer "Tell me when my rate hits X" on the rate, metro and quote pages: an email or phone number, the state, the loan type and a target rate. Send one alert when the target is hit, with a pre-filled quote link, plus a short monthly rate note. Hand alerted users to the voice agent when they reply.

    It captures early-stage buyers and homeowners waiting to refinance at almost no cost, and turns falling rates into a pipeline that converts on its own timing.

    Risks and how to handle them

    • Rates don't fall during the test

      Send the monthly rate note so the list stays warm.

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Competitor researchOwn Up: "Shop for your mortgage and save thousands"; customers "could save an average of $30,000 over the life of their loan", with a linked methodology; 4.9 from 4,690 verified reviews; free, "No credit impact"; a section for Realtors; says it is not a lender.Source
  • 08

    The first engineering hire is asked to "extend the AI voice agent to qualify more leads" and "improve close rates with better rate quotes". With two people, a human can't call every quote fast; an agent can, and it can open with the borrower's own numbers.

    High impactMedium effort First result in 1 month Software Engineer

    First action

    Measure today's time from quote to first contact.

    How to measure it

    Quote-to-application conversion

    Now
    –
    Target
    +25%
    By
    1 quarter

    Steps

    1. 1

      Baseline

      Time to first contact and quote-to-application rate today.

      Founder/CEO 1 day

    2. 2

      Consent

      Explicit opt-in to calls on the quote form, reviewed by counsel.

      Compliance counsel 1 week

    3. 3

      Trigger

      Quote completed or abandoned after the rate step.

      Software Engineer 3 days

    4. 4

      Script

      Open with the quote, answer top questions, hand off to a licensed loan officer.

      Software Engineer 1 week

    5. 5

      Log outcomes

      Reason codes for every call.

      Software Engineer 3 days

    6. 6

      Review calls

      Listen to ten calls a week and fix the script.

      Founder/CEO weekly

    First result
    1 month
    Cost
    Engineering time and per-minute voice costs
    Owner
    Software Engineer

    Smallest test

    Call back one week's abandoned quotes within five minutes and compare.

    Why it works

    When a quote is completed (or abandoned after the rate step), the voice agent calls within five minutes for borrowers who opted in, opens with their best rate and the gap to the median, answers the top questions (points, lock, closing time), and books a licensed loan officer or starts the application. Log every call outcome to learn which questions block applications.

    Speed to first contact is one of the strongest drivers of mortgage conversion, and a call that starts with the borrower's own quote feels like service, not sales.

    Risks and how to handle them

    • Unwanted calls

      Call only borrowers who opted in, and offer text-first.

    • The agent gives advice it shouldn't

      Hand anything beyond the quote to a licensed loan officer.

    Evidence

    • HiringSoftware Engineer (San Francisco, in person; $100K-$200K plus 0.5%-3% equity; current team of 2): "Extend the AI voice agent to qualify more leads", "Improve close rates with better rate quotes" and "Automate loan processing with browser agents and automated underwriting"; founded January 2025, backed by Y Combinator and Olive Tree Capital.Source
    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
  • 09

    Purchase loans start with real estate agents, and both Own Up (a section for Realtors) and Better (an agent match service) court them. Chestnut's automation and instant quotes are exactly what an agent needs when a buyer wants to make an offer tonight.

    High impactMedium effort First result in 2 months Founder/CEO

    First action

    List 50 active buyer's agents in one Texas metro.

    How to measure it

    Agent-sourced applications

    Now
    –
    Target
    20 a month
    By
    2 quarters

    Steps

    1. 1

      Pick a metro

      Where quotes already come from.

      Founder/CEO 1 day

    2. 2

      Build the agent page

      Referral link, same-day pre-approval promise, status updates.

      Software Engineer 1 week

    3. 3

      Status updates

      Automatic milestone texts or emails to the agent, with borrower consent.

      Software Engineer 2 weeks

    4. 4

      Compliance check

      No payments for referrals; disclosures reviewed by counsel.

      Compliance counsel 1 week

    5. 5

      Recruit agents

      Direct outreach and brokerage meetings.

      Founder/CEO ongoing

    6. 6

      Measure

      Applications and closings per agent.

      Founder/CEO monthly

    First result
    2 months
    Cost
    Engineering time and travel
    Owner
    Founder/CEO

    Smallest test

    Sign five agents in one metro and track their buyers for a month.

    Why it works

    Build a simple agent page and a dedicated number or link: the agent sends a buyer, the buyer gets a quote in minutes and a pre-approval letter the same day, and the agent gets status updates through to closing. Recruit the first 20 agents in one metro through direct outreach and local brokerage meetings.

    One active agent sends several buyers a year; fast, reliable pre-approvals and status updates are what agents judge lenders on.

    Risks and how to handle them

    • Agents stay loyal to existing loan officers

      Pitch as the backup for nights and weekends first.

    Evidence

    • Competitor researchBetter: "Simple, Online, AI-Powered Mortgage"; a "3 min | No credit impact" start; rate pages for purchase, refinance, cash-out, HELOC and VA; a HELOC calculator and a "HELOC vs. cash-out refinance" comparison; an agent match service for real estate agents.Source
    • Competitor researchOwn Up: "Shop for your mortgage and save thousands"; customers "could save an average of $30,000 over the life of their loan", with a linked methodology; 4.9 from 4,690 verified reviews; free, "No credit impact"; a section for Realtors; says it is not a lender.Source
    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
  • 10

    Chestnut sells cash-out refinances, home equity loans and HELOCs, and Better already has a HELOC calculator and a "HELOC vs. cash-out refinance" comparison. Homeowners who are happy with their existing rate are a large audience that a purchase-and-refi pitch misses.

    Medium impactMedium effort First result in 1 month Software Engineer

    First action

    Write the calculation and three example homeowners.

    How to measure it

    Home equity quotes

    Now
    –
    Target
    60 a month
    By
    1 quarter

    Steps

    1. 1

      Agree the math

      Blended cost of a HELOC plus first mortgage against a cash-out.

      Founder/CEO 1 day

    2. 2

      Build

      Inputs, comparison, quote buttons for both.

      Software Engineer 1 week

    3. 3

      Compliance review

      Assumptions and disclosures.

      Compliance counsel 3 days

    4. 4

      Publish

      On state, metro and rate pages.

      Software Engineer 1 day

    5. 5

      Promote

      To the rate-alert list and closed borrowers.

      Founder/CEO 1 week

    6. 6

      Measure

      Calculator uses to quotes to applications.

      Founder/CEO monthly

    First result
    1 month
    Cost
    Engineering time
    Owner
    Software Engineer

    Smallest test

    A static comparison table for three example homeowners.

    Why it works

    Build a calculator: current balance and rate, home value, cash needed. It shows the monthly cost and total interest of a HELOC against a cash-out refinance, then offers real quotes for both from the rate-shopping agent. Publish it on the Texas and Colorado pages.

    The calculator answers the homeowner's real question and puts them straight into the right product, with Chestnut able to quote either.

    Risks and how to handle them

    • Variable HELOC rates make comparisons misleading

      Show a rate-rise scenario next to the base case.

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Competitor researchBetter: "Simple, Online, AI-Powered Mortgage"; a "3 min | No credit impact" start; rate pages for purchase, refinance, cash-out, HELOC and VA; a HELOC calculator and a "HELOC vs. cash-out refinance" comparison; an agent match service for real estate agents.Source
  • 11

    Tomo advertises "98% on-time closing" against a 40% industry average, and a directory already repeats Chestnut's own claim of closings in 2 to 3 weeks. A young lender has few loans, which makes a complete, loan-by-loan record easy to publish and hard for big lenders to match.

    Medium impactLow effort First result in 1 quarter Founder/CEO

    First action

    Record promised and actual close dates for every loan so far.

    How to measure it

    Quote-to-application rate for visitors who see the record

    Now
    –
    Target
    +10%
    By
    2 quarters

    Steps

    1. 1

      Collect data

      Promised and actual close dates for every loan.

      Founder/CEO 1 day

    2. 2

      Define on time

      Closed on or before the contract date.

      Founder/CEO 1 hour

    3. 3

      Anonymise

      No names, addresses or exact amounts.

      Compliance counsel 2 days

    4. 4

      Publish the table

      Auto-update as loans close.

      Software Engineer 3 days

    5. 5

      Use it

      In agent outreach and next to the quote button.

      Founder/CEO ongoing

    6. 6

      Review misses

      Explain each late close and the fix.

      Founder/CEO monthly

    First result
    1 quarter
    Cost
    Staff time
    Owner
    Founder/CEO

    Smallest test

    Share the record with five agents and ask if it changes their view.

    Why it works

    Keep a public table: each closed loan (anonymised) with product, state, days from application to close, the date promised and whether it closed on time. Show the running on-time rate and median days to close at the top. Use it in agent outreach and the quote flow.

    Missed closings are what buyers and agents fear most. A full record, including misses, is more believable than a headline percentage.

    Risks and how to handle them

    • Early misses look bad

      Explain each miss; candour is the point of the page.

    Evidence

    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
    • Directory listingExtruct's profile of chestnutmortgage.com is titled "Chestnut Funding", lists "Instant quoting in under 2 minutes", 100+ lenders, closings in 2 to 3 weeks and typical savings of about 0.5%, but gives the address as "1234 Example Street, Los Angeles" and lists two directors whose names appear in the UK broker's reviews.Source
  • 12

    Chestnut already has rate-drop alerts and an agent that shops 100+ lenders. Every closed loan is a future refinance, and the borrower will refinance with whoever tells them first that the numbers work.

    Medium impactLow effort First result in When rates next fall Software Engineer

    First action

    Add rate-watch consent to the closing checklist.

    How to measure it

    Repeat-borrower refinances

    Now
    –
    Target
    20% of eligible borrowers refinance with Chestnut when alerted
    By
    12 months

    Steps

    1. 1

      Consent at close

      Opt-in to rate watch in the closing package.

      Compliance counsel 1 week

    2. 2

      Set break-even rule

      Saving must cover costs within a stated period.

      Founder/CEO 1 day

    3. 3

      Weekly check

      Run each enrolled loan through the rate-shopping agent.

      Software Engineer 1 week

    4. 4

      Write the alert

      Monthly saving, break-even, one-click start.

      Founder/CEO 1 day

    5. 5

      Pre-fill

      Reuse data on file to shorten the application.

      Software Engineer 1 week

    6. 6

      Track

      Alerts sent, applications, closings.

      Founder/CEO monthly

    First result
    When rates next fall
    Cost
    Engineering time
    Owner
    Software Engineer

    Smallest test

    Enrol all loans closed so far and run one weekly check.

    Why it works

    Enrol every closed borrower (with consent) in a rate watch: Chestnut checks offers for their loan each week and messages them when a refinance saves more than its cost within a set break-even period, with the exact monthly saving and a one-click application using the data already on file.

    It turns one closed loan into a relationship and makes Chestnut the first lender a past borrower hears from when rates fall.

    Risks and how to handle them

    • Too many alerts

      Alert only when the break-even rule is met.

    Builds on

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
  • 13

    Own Up shows 4.9 from 4,690 verified reviews and Tomo 5.0 from 70+ Zillow reviews, while searches for Chestnut's reviews turn up only a UK broker's Trustpilot page. Every loan Chestnut closes without a review is proof it doesn't get to show.

    High impactLow effort First result in 2 weeks Founder/CEO

    First action

    Write the request and send it to every borrower closed so far.

    How to measure it

    Google and Zillow reviews

    Now
    –
    Target
    25 reviews
    By
    2 quarters

    Steps

    1. 1

      Set up profiles

      Google Business Profile and Zillow lender profile.

      Founder/CEO 1 day

    2. 2

      Write the request

      Two lines and two links.

      Founder/CEO 1 hour

    3. 3

      Automate timing

      Send 2-5 days after closing.

      Software Engineer 1 day

    4. 4

      Ask past borrowers

      Everyone closed so far.

      Founder/CEO 1 day

    5. 5

      Display

      Review widget on the site and in outreach.

      Software Engineer 1 day

    6. 6

      Reply

      Answer every review within a week.

      Founder/CEO weekly

    First result
    2 weeks
    Cost
    $0
    Owner
    Founder/CEO

    Smallest test

    Ask the last five borrowers this week.

    Why it works

    Send a short review request two to five days after closing, when the borrower is happiest, with direct links to Google and Zillow. Show the reviews on the site and in agent outreach. Reply to every review, including critical ones.

    Reviews answer a borrower's first question about a new online lender (is it real and does it close?) and help the brand outrank the UK firm.

    Risks and how to handle them

    • Few closed loans

      Ask every one; ten genuine reviews are enough to start.

    Builds on

    Evidence

    • Search resultsSearching "Chestnut Mortgage" reviews returns mostly Trustpilot pages for Chestnut Mortgages, a UK mortgage broker at chestnutmortgages.co.uk (4.8 from 46 reviews); no US review profile for Chestnut appeared.Source
    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
    • Competitor researchOwn Up: "Shop for your mortgage and save thousands"; customers "could save an average of $30,000 over the life of their loan", with a linked methodology; 4.9 from 4,690 verified reviews; free, "No credit impact"; a section for Realtors; says it is not a lender.Source
  • 14

    Chestnut's pitch is that AI does "99% of human work", with Ashley as the AI loan officer assistant, and the launch demo video shows the product. Borrowers handing over income and credit data want to know what the AI decides and what a licensed person checks.

    Medium impactMedium effort First result in 1 month Founder/CEO

    First action

    Pick a recent loan and get the borrower's consent.

    How to measure it

    Walkthrough viewer conversion

    Now
    –
    Target
    +15% against non-viewers
    By
    1 quarter

    Steps

    1. 1

      Pick a loan

      Typical purchase or refinance, with consent.

      Founder/CEO 1 day

    2. 2

      Map the steps

      AI tasks, times, human checkpoints.

      Software Engineer 2 days

    3. 3

      Anonymise

      Remove all personal and property details.

      Compliance counsel 2 days

    4. 4

      Write the page

      Timeline with plain explanations.

      Founder/CEO 3 days

    5. 5

      Record the video

      Two minutes, screen and voice-over.

      Founder/CEO 2 days

    6. 6

      Place and measure

      Link from quote flow; compare conversion.

      Software Engineer 1 month

    First result
    1 month
    Cost
    Staff time
    Owner
    Founder/CEO

    Smallest test

    Share the walkthrough with the next ten quoted borrowers.

    Why it works

    Publish a step-by-step walkthrough of one real loan (with consent, anonymised): each task Ashley and the agents did (document review, rate shopping, conditions), how long each took, and every point where a licensed loan officer reviewed or approved. Turn it into a two-minute video and a page linked from the quote flow.

    It makes "AI lender" concrete and safe-feeling, answers the most common objection before the call, and is content no traditional lender can copy.

    Risks and how to handle them

    • Reveals process details competitors copy

      Show the borrower's view, not the internals.

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Launch videoThe launch post links a demo video titled "Chestnut: the 1st AI mortgage lender".Source
    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
  • 15

    Chestnut's agents already read mortgage documents and shop 100+ lenders, and the engineering brief includes automated processing with browser agents. The borrowers most ready to switch are the ones already holding a Loan Estimate from another lender, but today they would have to start a quote from scratch.

    High impactMedium effort First result in 6 weeks Software Engineer

    First action

    Test reading accuracy on ten sample Loan Estimates.

    How to measure it

    Upload-to-application rate

    Now
    –
    Target
    25%
    By
    1 quarter

    Steps

    1. 1

      Test extraction

      Rate, points, fees and loan terms from sample estimates.

      Software Engineer 1 week

    2. 2

      Match the scenario

      Run the same loan through the rate-shopping agent.

      Software Engineer 1 week

    3. 3

      Design the comparison

      Line by line, with total cost over five years.

      Founder/CEO 3 days

    4. 4

      Privacy and compliance

      Consent, data retention and disclosure review.

      Compliance counsel 1 week

    5. 5

      Launch

      Upload page linked from rate and metro pages.

      Software Engineer 1 week

    6. 6

      Promote to agents

      Agents send buyers who want a check.

      Founder/CEO ongoing

    7. 7

      Measure

      Uploads, beats, applications.

      Founder/CEO weekly

    First result
    6 weeks
    Cost
    Engineering time
    Owner
    Software Engineer

    Smallest test

    Offer manual second opinions by email for two weeks and count requests.

    Why it works

    Let Texas and Colorado borrowers upload a Loan Estimate they've received. Chestnut reads the rate, points and fees, runs the same scenario across its lenders and shows, line by line, where it can beat the offer, or says honestly that the offer is good. A beat converts straight into an application with the data already filled in.

    It meets borrowers at their moment of decision, uses the document reading Chestnut has already built, and gives a concrete, dollar-level reason to switch.

    Risks and how to handle them

    • Mistakes in reading the document

      Show extracted values for the borrower to confirm before comparing.

    • Sensitive data

      Short retention, clear consent, no use beyond the comparison without opt-in.

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • HiringSoftware Engineer (San Francisco, in person; $100K-$200K plus 0.5%-3% equity; current team of 2): "Extend the AI voice agent to qualify more leads", "Improve close rates with better rate quotes" and "Automate loan processing with browser agents and automated underwriting"; founded January 2025, backed by Y Combinator and Olive Tree Capital.Source
  • 16

    Chestnut is a San Francisco company in the YC network, but it can only lend in Texas and Colorado. Its home audience of startup and tech employees includes people relocating or buying second homes in Austin, Dallas and Denver, and they already trust YC companies and AI-first products.

    Medium impactLow effort First result in 1 month Founder/CEO

    First action

    Count how many quotes so far came from people moving from out of state.

    How to measure it

    Relocation quotes

    Now
    –
    Target
    40 a month
    By
    1 quarter

    Steps

    1. 1

      Check the data

      Share of quotes from out-of-state movers.

      Founder/CEO 1 day

    2. 2

      Write the page

      Stock compensation, remote income and timing questions answered.

      Founder/CEO 3 days

    3. 3

      Compliance check

      Licensing statement: Texas and Colorado properties only.

      Compliance counsel 2 days

    4. 4

      Seed in the network

      YC network, founder communities, startup perks lists.

      Founder/CEO 2 weeks

    5. 5

      Partner with relocation newsletters

      One sponsored mention each in Austin and Denver.

      Founder/CEO 1 month

    6. 6

      Compare

      Conversion against search traffic.

      Founder/CEO monthly

    First result
    1 month
    Cost
    Under $2,000 in sponsorships
    Owner
    Founder/CEO

    Smallest test

    One post in YC's network offering relocation quotes.

    Why it works

    Test a relocation offer for this audience: a page for people moving to Texas or Colorado (equity-heavy income, stock compensation and remote-work questions answered), shared through YC's network, startup communities, relocation newsletters and company perks programs. Measure whether these borrowers convert better than search traffic.

    It uses the network Chestnut already has, and its AI-first pitch fits this audience better than most.

    Risks and how to handle them

    • Small audience

      Keep it a cheap test and stop if quotes stay low.

    Evidence

    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
    • Public recordsSAM.gov registration: DETTA, INC. (dba CHESTNUT MORTGAGE), San Francisco, a Delaware corporation, active since May 6, 2026, registered for federal assistance awards.Source

Big bets 3

Larger plays that can change the trajectory.

  • 17

    Chestnut is licensed in two states while Tomo has rate pages for 40 states plus DC. Each new licence costs time and money, and once the site is indexed it will attract quote requests from states Chestnut can't serve yet.

    High impactHigh effort First result in 6-12 months Founder/CEO

    First action

    Add a "not in your state yet" waitlist to the quote flow.

    How to measure it

    Day-one applications in the new state

    Now
    –
    Target
    50 in the first month
    By
    12 months

    Steps

    1. 1

      Add the waitlist

      State, loan type, timing.

      Software Engineer 2 days

    2. 2

      Collect for a quarter

      Track sign-ups by state.

      Founder/CEO 3 months

    3. 3

      Cost each state

      Licensing time, fees and requirements, with counsel.

      Compliance counsel 1 month

    4. 4

      Rank

      Demand and loan size against cost.

      Founder/CEO 1 week

    5. 5

      License

      Apply in the top state.

      Compliance counsel 3-6 months

    6. 6

      Launch to the list

      Email the waitlist the day it opens.

      Founder/CEO 1 day

    First result
    6-12 months
    Cost
    Licensing fees and counsel time
    Owner
    Founder/CEO

    Smallest test

    Run the waitlist for 30 days and look at the top three states.

    Why it works

    Let out-of-state visitors join a waitlist by state, with their loan type and timing. Combine waitlist size, average loan size and licensing cost and time per state into a ranked list, and license the top state. Tell the waitlist the day it opens.

    It turns wasted out-of-state traffic into a launch list and makes the next licence a data decision with buyers ready on day one.

    Risks and how to handle them

    • Waitlist goes stale

      Send a quarterly update and the rate note.

    Builds on

    Evidence

    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
  • 18

    Chestnut says manual work costs "$12,000+ per loan" and that it is "starting as a broker" while it automates origination and underwriting. Tomo already leads with zero lender fees. If automation cuts Chestnut's cost per loan, the clearest way to show it is a price borrowers can compare at a glance.

    High impactHigh effort First result in 6-12 months Founder/CEO

    First action

    Start measuring cost per closed loan, by step.

    How to measure it

    Win rate after the fee change

    Now
    –
    Target
    +20%
    By
    12 months

    Steps

    1. 1

      Measure cost

      Staff and vendor cost per closed loan, by step.

      Founder/CEO 1 month

    2. 2

      Automate the costliest step

      Pick the biggest cost and automate it first.

      Software Engineer 1 quarter

    3. 3

      Model pricing

      No fee, flat fee and current pricing, by loan size.

      Founder/CEO 2 weeks

    4. 4

      Compliance review

      Fee disclosures and advertising.

      Compliance counsel 2 weeks

    5. 5

      Launch

      On rate pages, quotes and the methodology page.

      Software Engineer 1 week

    6. 6

      Track

      Win rate and margin per loan.

      Founder/CEO monthly

    First result
    6-12 months
    Cost
    Fee revenue given up per loan, offset by volume
    Owner
    Founder/CEO

    Smallest test

    Waive the lender fee for one month in one metro and compare win rates.

    Why it works

    Track the real cost per closed loan as automation ships. When it falls below a set level, launch a simple price: no lender fee (or one flat fee) on every loan, shown next to the rate. Use it as the headline for the shift from broker to lender and as the proof that the AI does the work.

    Borrowers understand "no lender fee" instantly; it turns an internal efficiency gain into a public advantage.

    Risks and how to handle them

    • Margin too thin

      Start with a flat fee and remove it only when cost per loan supports it.

    Builds on

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Competitor researchTomo Mortgage: "Low Rates, No Gotchas"; rates about 0.5% below the industry average; zero lender fees; rates published online; "98% on-time closing"; state rate pages for 40 states plus DC, including Texas and Colorado; 5.0 from 70+ Zillow reviews; Bankrate "2025 Best Online Lender".Source
    • HiringSoftware Engineer (San Francisco, in person; $100K-$200K plus 0.5%-3% equity; current team of 2): "Extend the AI voice agent to qualify more leads", "Improve close rates with better rate quotes" and "Automate loan processing with browser agents and automated underwriting"; founded January 2025, backed by Y Combinator and Olive Tree Capital.Source
  • 19

    The founder previously built a loan origination system that powered about 2% of US mortgages a year, and Chestnut is building agents that automate document review, processing and underwriting. Small lenders and credit unions face the same manual costs Chestnut's launch post describes but can't build agents themselves.

    High impactHigh effort First result in 6 months Founder/CEO

    First action

    Ask three lenders on the panel what their most manual step costs them.

    How to measure it

    Lender partners live

    Now
    –
    Target
    3
    By
    12 months

    Steps

    1. 1

      Interview lenders

      Costliest manual steps at five small lenders.

      Founder/CEO 1 month

    2. 2

      Pick one module

      The step Chestnut automates most reliably.

      Founder/CEO 1 week

    3. 3

      Harden it

      Security, audit trail and error handling for outside use.

      Software Engineer 1 quarter

    4. 4

      Price per loan

      Below the partner's manual cost.

      Founder/CEO 1 week

    5. 5

      Pilot

      One lender, 50 loans.

      Founder/CEO 1 quarter

    6. 6

      Decide

      Scale, keep narrow or stop.

      Founder/CEO 1 day

    First result
    6 months
    Cost
    Engineering time; partner security reviews
    Owner
    Founder/CEO

    Smallest test

    Run one partner's sample loan files through the module by hand.

    Why it works

    Package one proven automation (for example document review and conditions) as a service for a handful of small lenders or credit unions, priced per loan. Start with lenders already on Chestnut's rate-shopping panel, where the relationship exists. Keep it narrow so it funds and hardens the core product rather than distracting from it.

    It adds revenue that doesn't depend on rates or licences, and every partner's volume trains and tests the automation Chestnut uses itself.

    Risks and how to handle them

    • Distracts a tiny team from its own borrowers

      Cap it at one module and one pilot until the core funnel works.

    Evidence

    • Y Combinator profileChestnut describes itself as "The first AI mortgage lender": AI agents that "automate 99% of human work and save customers 0.5%+ in interest"; licensed in Texas and Colorado for purchases, refinances and cash-outs; founded 2025, Y Combinator Spring 2025 batch, team size 2, San Francisco.Source
    • HiringSoftware Engineer (San Francisco, in person; $100K-$200K plus 0.5%-3% equity; current team of 2): "Extend the AI voice agent to qualify more leads", "Improve close rates with better rate quotes" and "Automate loan processing with browser agents and automated underwriting"; founded January 2025, backed by Y Combinator and Olive Tree Capital.Source
    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source

Wildcards 2

Unexpected experiments, each with a cheap way to test it.

  • 20

    Chestnut's rate-shopping agent sees the whole spread of offers across 100+ lenders for identical borrowers, data most lenders and publishers don't have. Own Up builds its pitch on "save thousands" by shopping; Chestnut can show the size of the gap every week.

    Medium impactMedium effort First result in 1-2 months Founder/CEO

    First action

    Define three borrower profiles and pull this week's spread.

    How to measure it

    Subscribers and mentions

    Now
    –
    Target
    1,000 subscribers and 3 press mentions
    By
    2 quarters

    Steps

    1. 1

      Define profiles

      Three standard borrowers per state.

      Founder/CEO 1 day

    2. 2

      Automate the pull

      Best, median, worst offer weekly.

      Software Engineer 1 week

    3. 3

      Compliance review

      Wording and disclosures, without naming lenders.

      Compliance counsel 1 week

    4. 4

      Publish

      Archive page plus email.

      Founder/CEO weekly

    5. 5

      Pitch

      Local real estate reporters in four metros.

      Founder/CEO monthly

    6. 6

      Measure

      Subscribers, links, quotes from the report.

      Founder/CEO monthly

    First result
    1-2 months
    Cost
    Staff time
    Owner
    Founder/CEO

    Smallest test

    Publish one report and pitch it to two reporters.

    Why it works

    Publish a short weekly report: for three standard Texas and Colorado borrower profiles, the best, median and worst offer found, and what the gap means in dollars over five years. Send it to the rate-alert list, local real estate reporters and agents, and keep an archive page.

    "Shopping saves you $X this week" is a story reporters and agents reuse, and it promotes Chestnut's core product without an ad.

    Risks and how to handle them

    • Lenders on the panel object

      Publish anonymised ranges only, never lender names.

    Evidence

    • Launch postLaunch YC post "Chestnut, the first AI mortgage lender": Ashley, an "AI loan officer assistant"; a rate-shopping agent that compares rates across 100+ lenders, with Chestnut "starting as a broker"; origination automation; rate-drop alerts; purchases, refinances, home equity loans and HELOCs in Texas and Colorado; the call to action is an instant quote on the website; it says manual work costs lenders and borrowers "$12,000+ per loan".Source
    • Competitor researchOwn Up: "Shop for your mortgage and save thousands"; customers "could save an average of $30,000 over the life of their loan", with a linked methodology; 4.9 from 4,690 verified reviews; free, "No credit impact"; a section for Realtors; says it is not a lender.Source
  • 21

    Directories already describe Chestnut from partial information, one with a placeholder address, because nothing on chestnutmortgage.com is readable by search engines or AI assistants. When people ask an assistant "which lender should I use in Austin?", answers are built from pages like these.

    Medium impactMedium effort First result in 2-3 months Founder/CEO

    First action

    Export the 30 most common borrower questions from Ashley's conversations.

    How to measure it

    FAQ-sourced quotes

    Now
    –
    Target
    50 a month
    By
    2 quarters

    Steps

    1. 1

      Export questions

      Top 30 from chat and voice logs, anonymised.

      Software Engineer 2 days

    2. 2

      Write answers

      Short, factual, state-specific where needed.

      Founder/CEO 3 weeks

    3. 3

      Licensed review

      A licensed loan officer reviews and signs each answer.

      Licensed loan officer 1 week

    4. 4

      Publish with markup

      FAQ markup, dates and quote links.

      Software Engineer 2 days

    5. 5

      Refresh

      Update when rules or rates change.

      Founder/CEO quarterly

    6. 6

      Measure

      Visits, citations and quotes.

      Founder/CEO monthly

    First result
    2-3 months
    Cost
    Writing time
    Owner
    Founder/CEO

    Smallest test

    Publish ten answers and track impressions for a month.

    Why it works

    Once the site is open to crawlers, publish 30 short, factual answers to the questions Ashley and the voice agent hear most from Texas and Colorado borrowers (points vs rate, HELOC vs cash-out, closing timelines, state-specific steps), each with a date, a licensed reviewer and a link to a quote. Mark them up as FAQs and keep them current.

    Chestnut has the real question log; turning it into clear answers makes the site the source assistants and searchers cite.

    Risks and how to handle them

    • Answers go out of date

      Date every answer and review quarterly.

    Builds on

    Evidence

    • Your websiteFetches of https://chestnutmortgage.com/, https://www.chestnutmortgage.com/, http://chestnutmortgage.com/ and /robots.txt all returned HTTP 403 on November 10, 2026; a site: search for chestnutmortgage.com returned no pages from the site itself, only third-party directories and Y Combinator pages.Source
    • Directory listingExtruct's profile of chestnutmortgage.com is titled "Chestnut Funding", lists "Instant quoting in under 2 minutes", 100+ lenders, closings in 2 to 3 weeks and typical savings of about 0.5%, but gives the address as "1234 Example Street, Los Angeles" and lists two directors whose names appear in the UK broker's reviews.Source
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Second Front Systems growth strategyA public-benefit software company whose 2F Suite (Game Warden, Workshop and Frontier) lets commercial software companies build, authorize…22 growth ideas
  • Foundation Industries growth strategyA domestic CNC machining partner for hard-tech companies: approved customers upload parts to a portal, get quotes, pay and track…22 growth ideas
  • Godela growth strategyAn AI physics engine for engineers: it trains physics-constrained models on a team's existing simulation, CAD and test data, so engineers…23 growth ideas
  • Harochi growth strategyA Berlin GTM engineering firm that builds, runs and then hands over outbound and revenue systems: signal-based targeting, Clay enrichment…23 growth ideas
  • Chai Discovery growth strategyFrontier AI models (Chai-1, Chai-2, Chai-3) and a computational workspace for designing molecules, especially antibodies and miniproteins…21 growth ideas
  • Mirai Clinical growth strategyA Nevada body-care brand, seen on Shark Tank, making Japanese persimmon soaps, body washes, shampoos, wipes, deodorant and laundry and…20 growth ideas
  • Modulr growth strategyA regulated payments automation platform for payroll, supplier payments, spend management and collections, with accounts, pay-outs…22 growth ideas
  • BILL growth strategyA financial operations platform for small and midsize businesses and accounting firms: accounts payable and receivable, Spend & Expense…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 10, 2026, and cites them throughout. Chestnut (Detta, Inc. dba Chestnut Mortgage) did not commission, review or endorse it, and Programmable is not affiliated with Chestnut (Detta, Inc. dba Chestnut Mortgage). Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Chestnut (Detta, Inc. dba Chestnut Mortgage) runs its business.

Something wrong, or do you represent Chestnut (Detta, Inc. dba Chestnut Mortgage)? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 305 ideas

    Let Google in, own the brand name, fix wrong directory listings, publish sample rates and prove the 0.5% claim.

  2. Days 31 to 6010 ideas

    Metro pages, rate alerts as sign-up, five-minute voice callbacks, the agent pre-approval line, the HELOC calculator, rate watch for closed borrowers, reviews, the Ashley walkthrough, the Loan Estimate second opinion and the relocation test.

  3. Days 61 to 906 ideas

    The on-time closing record, the next-state decision, the no-lender-fee price, licensing the automation, the spread report and the public FAQ.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value A lower rate from shopping many lenders at once. The most cited problem: overpaying by not shopping lenders.

  • An instant quote with no credit impact
  • Lender fees and surprises at closing
  • In their words: “rate”

Market

Competitive landscape

3 competitors compared: Own Up, Tomo Mortgage and Better.

  • Gap to exploit: Proof, not just a claim, that AI shopping across 100+ lenders lowers the rate
  • Own Up: Hands borrowers to other lenders instead of closing the loan itself
  • Tomo Mortgage: Same 0.5% savings message as Chestnut, so the claims need proof to stand apart

Hiring signals

Chestnut is a team of two hiring its first Software Engineer (San Francisco, in person).

  • Software Engineer: Voice agent lead qualification, better rate quotes, automated underwriting. Ideas 4, 8, 15, 18, 19.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks firstNew accounts get 100,000 free enrichment credits with code BETA100K (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammableMediaAn AI note taker, interpreter and webinar host for your calls (opens programmable.media in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence