On September 16, I gave the AI keynote at ML Mortgage’s Annual Sales Summit in Claremont, California. About forty loan officers and branch managers were in the room. They ranged in age from 19 to 67, most were millennials, and mine was the first session of the day’s AI block. The talk was called Don’t Automate the Handshake.
This is the written version. It covers what I argued, the prompts we ran live, the two guardrails I’d give any sales team in a regulated industry, and what I’d tell anyone thinking about booking an AI speaker for their own team. If you just want the deck and every prompt in one place, it’s all on the free resource page for the talk. It doesn’t ask for your email.
Why I didn’t give a tool tour
Most AI talks for sales teams are a tour of apps, and that tour goes stale in about ninety days. The tools change every quarter. The jobs people need them for don’t. So I built the hour around jobs rather than products, and I made it about mortgage from the first slide instead of putting a generic deck in front of a mortgage audience.
The argument: polish stopped being a differentiator
AI made competent output free. Anyone in that room can now write a clean follow-up email, a clear explanation of a rate, or a polished market update in seconds. So can every competitor. When everyone’s floor rises at the same time, effort stops showing.
So what’s left? Judgment, and the relationship. There is no tool for sitting across from somebody on the worst day of their loan. The whole talk hangs on that line: automate the reach, never the relationship.
I also described AI as a Swiss Army knife. It does a surprising number of things reasonably well and none of them best. That sets the right expectation for a sales team: it’s a fast assistant with no license, and it never tells you when it’s guessing.
The three jobs AI is good at in a sales role
Everything worth doing with AI right now fits into one of three jobs.
1. Memory
Your database is the most valuable thing you own, and it’s the thing you touch least. AI is good at showing you who you’ve forgotten: the buyer you pre-approved who went quiet, the past clients now well above today’s rate, the emails still waiting on a reply from you.
2. Translation
You already know the answer. AI helps you say it the way this particular person needs to hear it. One prompt we ran live asked it to explain a 7-year ARM in terms the borrower already understands. The room picked food. The answer described a restaurant that locks your menu price for seven years, then follows the market, with a cap on how much the price can jump in any one year. The facts didn’t change, but they got easier to hear.
3. Prep
This is the least glamorous of the three, and it’s the one that moves the number. Before a hard call, describe the scenario and ask for the five toughest questions the borrower is likely to ask. AI doesn’t know more than you do. It just isn’t nervous.
Two guardrails for a regulated industry
Mortgage is regulated. Leave a sales team to experiment on its own and you’ll end up with forty different AI policies, all made up on the spot. Two rules cover most of the risk.
Nothing AI writes reaches a borrower until a human has read every word. Ask it about a loan program that doesn’t exist and it will describe one, with total confidence. You hold the license. It doesn’t.
Describe the situation, never the person. “Self-employed borrower, two years of returns, tight DTI, 15% down” gets you the same help as a name, a paystub and an account number, and it keeps anyone’s personal information out of the tool.
For branch managers, I left a four-line starter policy on the resource page. The fast branch isn’t the one with no rules. It’s the one where nobody has to wonder what the rules are.
What happened in the room
We ran two prompts live, with everyone’s phones out. The first asked for three short texts to a pre-approved buyer who’d gone quiet for three weeks. Then I asked everyone to add the one sentence only they could write. That second step is the point: the AI drafts the reach, and the person supplies the relationship.
The age range mattered. People from 19 to 67 don’t share a starting point, so every example was something anyone could do on their own phone, with a free tool, in under a minute, before the session ended.
Afterward, branch managers asked about bringing a version of the session to their own teams and sister branches. That’s the response I was hoping for. Not “cool demo,” but “my people need this.”
Someone in the audience filmed the talk on a phone:
If you’re thinking about booking an AI speaker for your team
Having built this for a regulated sales audience, here’s what I’d ask any AI speaker before booking them, me included:
- Have they built anything? Ask what they’ve shipped. People who build with these tools every week give different advice than people who only read about them.
- Will it be about your industry? A generic deck costs your team an hour. The examples should use your products, your compliance rules and your customers.
- Is anything live? People learn more by running a prompt on their own phone than by watching someone else’s screenshots.
- What will people leave with? Everyone should walk out with prompts they can use on Monday and a clear rule for what not to do.
- Does it cover risk? In mortgage, insurance, financial services, healthcare or legal, a talk that skips guardrails is a liability.
Common questions
How long is the keynote?
Usually 45 to 60 minutes, including Q&A. The ML Mortgage session was about 45 minutes, followed by questions.
Can the talk be adapted for another industry?
Yes. The structure (three jobs, two guardrails, one thing that stays human) works for any team that sells or serves clients. The examples, the prompts and the compliance points are rebuilt for your industry every time.
Do you offer hands-on workshops as well as keynotes?
Yes. I offer half-day and full-day workshops, plus ongoing team training for companies that want people practicing, not just listening.
Bring this talk to your team
I give AI keynotes, hands-on workshops and ongoing training for sales teams, built around your industry and your agenda. I spent eleven years at Apple teaching people how to use technology, and these days I spend my weeks building AI systems for businesses. I travel from Las Vegas.
If you’re planning a sales summit, a kickoff or a branch meeting, see how I work or get in touch. And if you were in the room in Claremont, everything from that hour is on the Don’t Automate the Handshake page.
Leave a Reply