AI CRM for Loan Officers Compared: BNTouch, Shape, Total Expert

There is no single winner. BNTouch publishes the most (from $95 per user per month, mortgage-only, LOS sync, a lead-grading assistant). Shape and Total Expert price by quote, so pick by job: BNTouch for a priced, mortgage-built CRM, Shape for AI calling and texting, Total Expert for larger lender teams.

That answer holds only until you test three things the vendors' pages do not settle: whether loan-stage data syncs both ways with your LOS, whether a person approves AI messages before they send, and whether a lead score comes with reasons you could explain to a compliance reviewer. Prices can be read off a page; those three things have to be watched in a live demo.

Follow me on Instagram@sagnikteaches

BNTouch, Shape and Total Expert side by side

The table marks what each vendor's own pages say. Where a page is silent, the cell says so, because silence is a question for the demo, not a tick.

Connect on LinkedInSagnik Bhattacharya

Everything in the table is the vendor's own description, which is marketing, not a test result. For a wider framework on picking a mortgage CRM, choosing a mortgage CRM with AI built in covers what matters beyond these three.

Subscribe on YouTube@sagnikteaches

What each vendor's price page tells you

BNTouch: the only one with numbers

BNTouch lists BNTouch CRM at $95 per user per month for two to ten users, $165 a month for a single user, and custom pricing above ten users. Its Mortgage AI page adds a one-time activation fee covering onboarding, and does not state the amount. Included in the listed plan: 5,000 records per user (extra records cost $5 per additional 1,000), 400-plus pre-built campaigns, a point-of-sale tool with an AI-assisted online 1003, and "Mortgage AI". Texting and voice include 100 SMS or MMS segments or minutes each month, with extra at $0.03 each. The standard plan is month to month with no contract, and a 12-month prepayment earns the thirteenth month free.

Two parts are extra. BNTouch's New Loan Opportunity Mining module lists Credit Check Alerts and Equity Alerts as excluded and at an extra fee, while Rate Refinance Miner, Strike Rate Alerts, Just Listed Alerts and Listed Alerts by Partner come with the CRM subscription. The page notes that credit alerts are available only where permitted by law and its data provider agreements. A managed White Glove service, where BNTouch runs the CRM for you, is also priced separately, on a call.

Shape: a quote, with a vendor-cited figure

Shape's pricing page has three tiers and no dollar amounts. Core is priced per user and covers lead management, workflow automation, lead routing, integrations, a mobile app, calendar and tasks, drip campaigns and built-in reports. Complete includes five users and adds an AI Calling Bot, AI Insights, landing pages, social tools, a customer portal and more. Enterprise is custom. An "AI Suite" add-on (AI calling, texting, insights, content and lead scoring) can be bolted on to Core. Calling, texting and email run on a preloaded balance at pass-through rates, or on an Unlimited Calling Plan. Shape's own comparison post on loan-officer CRMs, published in July 2026, gives a starting price of $119 per user per month; because that comes from a marketing post rather than the pricing page, confirm it in a written quote. The pricing page does say annual payment saves up to 20 per cent.

Total Expert: built for larger lenders, quote only

Total Expert's site describes a platform for financial institutions with an AI Sales Assistant "trained specifically on mortgage use cases", plus AI SMS and Scenario assistants and Customer IQ, which it describes as adding intent signals, enriched profiles and consent awareness. Its site has no pricing page, and third-party figures conflict, so none is quoted here. The practical step is to ask for a quote that names the seat price, any platform fee, the Encompass or other LOS connection, and whether AI assistants are priced separately or by usage.

A cost sum for a team of six

Take a small branch with five loan officers and one processor, six users, a database of 22,000 contacts and an expected 2,400 texts a month. The team and figures are invented, but the arithmetic uses BNTouch's published terms.

• Seats. 6 × $95 = $570 a month, $6,840 a year, plus the one-time activation fee (ask for the amount).
• Records. 6 × 5,000 = 30,000 included, so the 22,000 contacts fit with no extra charge.
• Texting. If the 100 included segments are per account, extra = (2,400 − 100) × $0.03 = $69 a month. If they are per user, extra = (2,400 − 600) × $0.03 = $54 a month. The page does not make plain which applies, so ask. That puts texting at roughly $650 to $830 a year.
• Opportunity Mining credit alerts. Extra fee, quote on a call.

So the published parts come to roughly $570 + $54 to $69 = $624 to $639 a month before activation, credit alerts and any managed service. For Shape and Total Expert there is no sum to run until the quotes arrive; the exercise is to send each of them the same one-page brief (six users, 22,000 contacts, 2,400 texts a month, your LOS, AI features wanted) and compare line by line. Our tutorial on what it costs to set up an AI CRM for a small team lists the costs that sit outside the licence.

LOS sync: three tests that separate one-way from two-way

A CRM that only imports contacts is a contact list. A CRM that sees loan stage changes can trigger follow-up at the right moment: a document request when a loan enters processing, a congratulations message at funding, a review request after closing. BNTouch says its LOS syncs both ways, and lists the systems above. Shape's mortgage page claims native LOS integrations with real-time updates and milestone syncing. Total Expert's current pages did not confirm a list, so ask.

1. Which fields sync, and which direction? Ask for the field list: loan stage, milestone dates, loan amount, rate lock date, property address. Then ask which are read-only in the CRM.
2. How often? Real time, every few minutes or nightly? A nightly sync means a clear-to-close message can arrive a day late.
3. What happens when a field changes in both places? The vendor should name which system wins. Ask what the log shows when a sync fails.

A practical demo step: move one test loan through two stages in your LOS sandbox while the vendor watches, and see how long the CRM takes to reflect each change and which automation fires. If the vendor cannot do this live, treat the sync claim as unproven.

Review before send: why the vendor pages leave it open

None of the three pages I read states whether a loan officer approves an AI-drafted message before it goes out. BNTouch's AI content generator is described as creating emails and texts in seconds using an integrated ChatGPT tool, which sounds like drafting for a person to use. Shape's agents are described as answering inbound calls, dialling out on aged leads, running text qualification and transferring live to a loan officer. Total Expert describes an assistant that nurtures leads through "compliant, human-like voice interactions", and its AI SMS Assistant page goes furthest on guardrails: consent is gated, quiet hours are enforced, opt-outs stop outreach instantly, and every message writes back to the contact record. Its Scenario Assistant produces priced refinance or purchase scenarios for you to review. Whether each message needs approval, and whether the setting can be locked by an administrator, is yours to ask.

Messages in mortgage work often carry regulatory wording: rates, terms and approval status are all areas where your compliance team will have rules. The CRM cannot decide those for you. What you can require is a control that makes your rules enforceable.

A sample failure and the control that prevents it

Say an AI-drafted text to a lead who has only completed an enquiry form reads: "Great news [first name], you're pre-approved! Rates are at 5.1% today." Two problems: the lead is not pre-approved, and a quoted rate may need required disclosures. If the setting is draft-and-approve, a loan officer deletes the text. If it auto-sends, it is out before anyone sees it. The control to ask for is an administrator-level switch that sets each template or campaign to "approve before send", with a log of who approved what and when.

For a deeper look at follow-up rules and compliance checks, see AI tools for loan officers: CRM, follow-up and compliance checks.

Lead scores you could explain to a reviewer

BNTouch says MAIA NextStep analyses loan data and past performance to grade leads and update mortgage records every hour. Shape lists lead scoring in its AI Suite. Total Expert describes intent signals and enriched profiles. None of the pages say how a grade or signal is explained. That matters for two reasons: you want to know which leads to call first, and you may need to explain a prioritisation decision later. Try the back-test below on every vendor you shortlist.

A 30-loan back-test

1. From last year, pick 30 leads: 15 that funded and 15 that dropped out. Strip names if the demo is on the vendor's system.
2. Load or describe their early-stage data: source, loan type, response times, credit band if you hold it.
3. Ask the vendor to score them and rank them, without showing outcomes.
4. Count how many of the top ten are funded loans, and how many of the bottom ten are drop-outs.
5. Ask for the reasons behind one high score and one low one.

An invented result: the top ten contain seven funded loans, the bottom ten contain six drop-outs. That is better than chance, which would put five funded in the top ten. If the vendor's score puts four funded loans in the top ten, it is no better than a coin toss on your data. And if it cannot give a reason for one score, you cannot explain it either. Thirty loans is a small sample, so treat the back-test as a screen for obvious failures, not a proof. Our tutorial on AI lead scoring for a small business explains how scoring works and where it breaks.

A new lead's first 24 hours under each approach

Feature lists blur together, so walk one lead through each product's described behaviour and note where a human acts. The timeline below is an illustration built from what the vendors say their tools do, not a recording of any product.

• Hour 0, a web form arrives at 9.40pm. In a CRM like BNTouch, the lead lands in the loan pipeline and a campaign or text may fire; MAIA NextStep may grade the lead. With Shape, the AI calling bot is described as able to dial out or text to qualify and transfer live to a loan officer. With Total Expert, the AI Sales Assistant is described as qualifying and nurturing through conversation.
• Hour 1. Ask in the demo: has anything been sent to the borrower yet, who approved it, and what does the record show?
• Morning. The loan officer opens the record. The test is whether the lead's answers, any call summary and the score with its reasons are all on one screen, so the first call starts from facts.
• Hour 24. If the lead has not replied, what does the sequence do next, and which step would a loan officer want to stop? Check that there is a visible pause button per lead.

The value of this walk-through is that it turns three marketing pages into the same set of observable events. Run it with each vendor, and write one line per hour. Where a vendor cannot answer "who approved it", note that as a gap, not a minor point.

What borrower data the AI sees, and where it goes

Mortgage records hold some of the most sensitive personal and financial detail a small business handles. Before any AI feature touches them, get written answers to four questions, and take the wording to your own compliance adviser, since the vendor cannot decide what your rules require.

1. Which fields does the AI read? Contact details only, or income, assets and credit information too? Ask for the list, and whether you can exclude fields.
2. Which AI provider processes them, and on what terms? Ask whether borrower data is used to train models, and for the vendor's data-processing terms in writing.
3. What is recorded of AI calls and texts? For voice agents in particular, ask whether calls are recorded, how consent is captured, and where recordings are stored and for how long.
4. What happens on exit? Ask how you export the contact data, notes, message history and any recordings, and in what format.

Consider a loan officer who tests an AI calling feature on a list of 40 past clients. Three clients had asked never to be called. The list was imported without the do-not-call flag, and the bot rang them. The cause was a field that did not map across, which is why the import test with known opt-outs matters. For call logging that feeds the CRM safely, see our tutorial on AI call summaries and logging enquiries in your CRM.

Fit by team type

• Solo loan officer. BNTouch's single-user price of $165 a month is the one figure you can test against your own time. If your lender already supplies a CRM, check what it lacks before paying for a second system. Shape's Core plan may suit if calling and texting are the main need, once you have a quote.
• Branch of five to ten. BNTouch's per-user rate applies up to ten users, so the sum above is directly usable. Add your LOS and approval tests to the demo list.
• Larger team or bank. Consent handling, multi-location controls and integration depth matter most. Total Expert's focus on financial institutions and Shape's Enterprise tier with branching are the ones to quote. Expect implementation work, which belongs in the cost, along with moving your data and retraining the team.
• Anyone whose data is messy. Clean it first. Duplicates and missing loan types distort every score and every sync; cleaning up a messy CRM with AI is a sensible first step.

A nine-question script for every CRM demo

1. Quote: seats, activation or set-up fee, platform fee, texting and calling rates, AI features priced separately?
2. LOS: which systems, which fields, which direction, how often? Move a test loan live.
3. AI messages: show the approve-before-send setting and who can change it.
4. Blocked words and required disclosure wording: show me.
5. Opt-out: reply "stop" and show every automation pausing.
6. Lead score: rank my 30-loan sample and give reasons for one high and one low.
7. Records: how many included per user, and the charge above that?
8. Contract: term, notice period, data export on exit, who owns the contact data.
9. Data: where is borrower information processed, and which AI provider handles it?

Checking it worked after sixty days

Pick three numbers before you start and read them again at day sixty: median time from enquiry to first reply, share of enquiries with a follow-up logged within a day, and loan officer hours a week spent on manual CRM updates. Compare with the sixty days before. Read ten AI-drafted messages a week and mark each as right, fixable or wrong; pause the automation if more than one in ten is wrong. Ask each loan officer whether they would object to switching it off. Where answers are lukewarm, the feature is probably adding clicks.

A realistic mistake shows why the sync-speed question matters. Picture a team whose LOS syncs to the CRM once a night. On Friday afternoon a closing is pushed back a week, but the CRM still holds the old date, so an automated reminder about Monday's closing goes to the borrower overnight. The borrower rings the loan officer, worried that something has gone wrong. The fix is to fire milestone messages only from fields that sync in real time, or to hold them for approval, and to test the timing with a sandbox loan before go-live.

Questions loan officers ask before a CRM demo

Which of the three publishes a price per user?

BNTouch does: plans start at $95 per user per month for two to ten users, or $165 a month for a single user, plus a one-time activation fee. Shape's pricing page says to get a quote, and Total Expert's site shows no price. Get written quotes from the other two before comparing.

Does any of them send AI messages without my approval?

The vendor pages I read do not say either way for text and email drafts. Shape describes AI agents that call and text leads and transfer live to a loan officer; Total Expert describes voice and SMS assistants that hold two-way conversations, with consent gating and quiet hours on SMS. Ask each vendor to show where approval is set and what is logged.

What is a lead score I can explain?

One where you can see the factors behind the grade and a compliance reviewer can understand them. Test by asking the vendor to show why one named lead scored high and another low. A score that is only a number or letter, with no reasons, is hard to defend if a borrower or auditor asks.

Do I need two-way LOS sync?

If loan stage changes should trigger follow-up, such as a clear-to-close message, yes. One-way import of contacts is much weaker. Ask which LOS fields sync, how often, and which direction. BNTouch lists Encompass, Calyx, LendingPad, MeridianMortgage and BytePro on its site.

Can a solo loan officer afford these?

BNTouch's single-user rate is published at $165 a month. A solo should price that against hours saved on follow-up before asking for quotes from the others. For a solo whose lender supplies a CRM already, the question is often whether the extras justify a second system.

Further reads

• AI CRM for Financial Advisers: What Is Built In and What Is Bolted On — The same built-in versus add-on test for advisory CRMs.
• What Is an AI CRM and Which Features Actually Save Time? — Which AI CRM features actually save time.
• AI Tools for Mortgage Lenders: Processing, Disclosures, Audit Trails — Disclosures and audit trails for lender AI tools.
• AI Loan Processing: Catch Missing Documents Before Underwriting — Catching missing documents before underwriting.
• HubSpot for Mortgage Brokers: Leads, Referrers and Case Stages — A general-purpose CRM set up for mortgage case stages.
• How to Hire a HubSpot or CRM Consultant to Set Up AI Features — When to hire help to configure a CRM's AI.
• Is It Worth Switching CRM Just for Better AI Features? — Whether moving to a new CRM for better AI pays back.
• Can Loan Officers Use AI Videos for Market Updates? Rules to Follow — Disclosure rules for AI avatar videos written before the first render, with a script template, caption checklist and review gate for loan officers.
• How Can AI Help Credit Unions? Member Service and Back-Office Jobs — Three jobs where AI saves a credit union real time, member enquiries, loan admin and call notes, with the guardrails, sample drafts and sums.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: BNTouch home, pricing and Mortgage AI pages; Shape home, pricing and mortgage pages and Shape's July 2026 comparison post on loan-officer CRMs; Total Expert home and AI Solutions pages (all fetched 1 November 2026). Total Expert publishes no price. Prices change, so confirm in a written quote.