Marketing and AI for SMEs Podcast: How Ruby He (CPA) Uses Data + AI to Drive Mortgage Leads

I sat down with Ruby He CPA and Founder of Ruby Strategic, for a practical conversation on AI adoption for small businesses. 

Ruby isn’t a theorist — she led AI implementation as General Manager of Operations for a 350-person company, and now consults mortgage brokers and SMEs on how to build AI solutions that actually work. 

If you’re a business owner trying to cut through the noise around AI, this one is worth your time. We cover what AI can realistically do for you, why your data is your most important asset, and what most implementations get wrong before they even start. 

You can check out our ads management services or client results if you’re looking for what we do at Digital Goliath.

• Watch on YouTube
• Listen on Apple Podcasts

Introduction

Jeremy: Ruby, welcome. Thanks for coming onto the show. I was intrigued when we spoke at our networking event. A lot of people are talking about AI and adoption — mostly in theory. When you told me about your background and what you’ve already built, I thought it would be worth getting into the practical side of it. Where can small businesses actually use it?

Ruby: Thank you for having me, Jeremy.

Jeremy: The way I like to run these is we build your credibility first — where you came from, why people should listen to you. And I like to get the pricing question out of the way early. What are you offering, and how do people get the first bite of the apple?

I feel like on a lot of shows, they leave that part so late that by the time they get there, the listener’s already thinking, “I can’t afford it anyway.” So let’s just get it out in the open.

Ruby: Sure. People can find me through my websites — rubystrategic.co and realaccounting.co. I offer a free 15-minute consultation. From there, a full consultation is $300 per hour. When clients engage me properly, I usually start with a proof of concept project for around $3,000. Then from there, we move into phased projects, which are typically around $30,000.

Jeremy: And who do we really want to be talking to today? If we can call out a target market.

Ruby: Mortgage brokers. Specifically, the owners of mortgage broking businesses — ideally with more than five staff members.

Jeremy: So in a typical mortgage broking business, the owner usually started as a broker themselves, got too busy, and started hiring. Is it a mix of local and offshore staff?

Ruby: Usually they’ll have a couple of brokers onshore in Australia, and then support staff in the Philippines, Nepal, or India. That’s the usual setup, and they grow from there.

Ruby’s Background

Jeremy: I want to get into your history because I think it’s what sets you apart. You were telling me you led an AI implementation as GM of operations — how many staff was that?

Ruby: The company was about 350 people, and I was the General Manager of Operations. I didn’t deal much with the sales and marketing side of things, but everything to do with operations — finance, customer service, HR, data, tech, IT — that was all under my remit.

Jeremy: All of those departments, all those different personalities. So your angle coming into AI consulting is really from an operations and implementation standpoint?

Ruby: Exactly. It’s from the leadership team’s point of view. As the leader in the business, I could see the pain points, I knew what problems needed solving, and I had the knowledge of AI to combine with that and provide a solution.

The First Implementation

Jeremy: The project you worked on there — was it about introducing AI?

Ruby: Yes. This was back in 2023, when ChatGPT came out and there was a lot of hype. Every business owner was panicking — “We’re going to be irrelevant in 12 months.” The first thing the leadership did was hire a Head of AI, which a lot of businesses did. But that person was very focused on the tech side of things, and the first implementation went nowhere. That became my problem, even though I didn’t have much of an AI tech background myself.

So I went back to first principles. What are the actual problems in this business that AI could solve? I learned the basics — natural language processing, like ChatGPT, and vision capabilities that can read documents.

Then I matched that against the business problems. We were a real estate company, and customers constantly asked: when is my property settling? When is my land settling? That was taking up a huge amount of our human customer service team’s time.

So I decided to get AI to solve that. I put together a team — AI developers, data people, and a portal developer to build the interface. I said, give me an MVP. They delivered a minimal viable product that could answer that one question.

We set up an enterprise-grade language model account, connected it to a data warehouse that held our property information, and built an interface for clients to interact with. The whole system let the AI answer clients based on actual property data — including the settlement date, which changes.

Jeremy: And with the previous Head of AI, it sounds like the gap was that he hadn’t learned the business well enough?

Ruby: That was the biggest obstacle. He just came in. He didn’t yet understand what the business actually did. That learning curve is long — I was hired in 2020, started as Head of Finance, and had to work through every part of the business before I knew where the bottlenecks were. That took years. He didn’t have that, and it showed.

The Purple Translator

Jeremy: So for your own consulting business, your value proposition is that you sit between the business side and the tech side?

Ruby: Exactly. A lot of businesses are missing that person who understands operations and understands tech. Business people are like red — passionate, driven. Tech people are like blue — very logical, very structured.

Jeremy: So you consult from the start, and you always begin with the business problems?

Ruby: Always. My consultation starts with the business problems. A lot of them are very similar, especially within the same niche — usually it’s customer service, repetitive tasks, time getting swallowed by things that don’t add value. I ask the business owner: where does your time go? What’s repetitive and not adding value?

When AI Isn’t the Right Starting Point

Jeremy: In your case study, you talked about guardrails on the AI — transparency so that the broker knows every interaction, and escalation the next day. I want to push on something. If a customer asks where their loan is and then starts asking follow-up questions — why is it taking so long, are my documents approved — I want to hear a voice. I want certainty. Do you think that works for a five-person team?

Ruby: For smaller businesses, I actually don’t always suggest using AI for direct client communication. Sometimes your brand is yourself, and that personal touch matters. If a business owner isn’t comfortable with AI engaging their clients, I don’t push it.

Instead, I look at the workflow. There’s a mortgage broking business in Melbourne we’re currently talking to — about 10 staff. The owner said, “I don’t feel comfortable with AI talking to my customers.” So I asked: what’s taking up the most time internally? He said compliance — checking payslips and bank statements.

So instead of natural language processing, we use AI vision — the optical side — to read documents and build logic to flag what doesn’t add up. Some people fake their payslips. We use AI to catch it.

Jeremy: That’s a different use case entirely, and it sidesteps the trust issue completely.

Custom Solutions vs. Off-the-Shelf Tools

Jeremy: Surely there are people building out-of-the-box solutions for mortgage brokers right now. Someone’s trying to be the authoritative AI tool for that industry. Is that what you’re aiming for?

Ruby: No. Building a product needs a lot of funding, and I’m not going down that path. Where I specialise is helping people make sense of the tools that already exist and choosing the right stack for their specific situation. There are already a lot of AI tools built for mortgage broking. I’m not trying to cover everything — but I can give them a consultation on: here’s your stack, here’s how you get the most out of it.

Jeremy: I genuinely think that’s where the most value is. In marketing, tools appear every day — it’s overwhelming for small business owners. Someone has to be on the other end, helping them figure out what’s worth using and what’s already been replaced. Your level of discernment and industry knowledge is what wins.

Ruby: Exactly. I’m not trying to build my own tools because there are already plenty out there. My job is to stay grounded in the business problem. We’re business people helping other business people solve business problems.

AI for Blue Collar Businesses

Jeremy: You work a lot in professional services. Can this work for blue collar at a certain size?

Ruby: Definitely. Tradies actually need more help, I think — they generally don’t have time to think about systems and software. But I look at it holistically. AI is the tip of the iceberg. Underneath it is your data, your systems, and your people. You have to look at them all together.

For blue collar businesses, they either don’t have a system, the system is messy, or they’ve signed up to a bunch of apps that are now all tangled up.

Jeremy: I see that all the time. Four or five vans on the road, owner is hands-on, lunches running over — how do you even start getting their data in order?

Ruby: I start with the CRM. Do they have one? If they don’t, I help them get one, and my team can help implement it. I don’t see myself as a consultant who just talks and hands over a plan. My people come in and help make it happen.

The Real Estate Agency Case Study

Jeremy: Can you give us a real example of how a business pulled this together — CRM, data warehouse, AI?

Ruby: Sure. We’re working with a real estate agency on the Gold Coast. They have a database of 50,000 people and their goal is to proactively reach out to find people who want to sell their house.

Their CRM is called Rex — a standard real estate software. We connected that to a data warehouse, cleaned the data, and then used AI to send text messages that have a natural conversation with prospective sellers. The message says something like: “We’re about to sell a house on your street — would you like to know what it goes for?”

From there, the AI reads the responses and assesses sentiment, scoring each reply so the human agent knows who to call first. The ones who are interested get prioritised. The ones who aren’t get handled without wasting a human’s time.

Jeremy: And all of that conversation history stays in your data warehouse, not a third party’s.

Ruby: That’s the point. If you buy a third-party tool, your data goes into their system. When I build a custom solution, all your data — the chat history, the sentiment scores, all of it — stays in your warehouse. That’s your moat. That’s the asset you’re building.

They had initially wanted to buy a third-party product called Rita, made by Cortell. I didn’t tell them not to buy it. But I said: if you want to future-proof your business and have AI that’s trained on your data, build something custom. That’s the difference.

People, Change Management, and Buy-In

Jeremy: The last thing I want to get into is the people side. A friend of mine who works in AI explained it to me clearly — your success with adoption depends entirely on who you’re talking to inside the business. If the key person thinks this is going to get them fired, they’re not going to help it work.

Ruby: You’re exactly right. AI implementation is not just about the tech and the tools — it’s about cultural change. It’s a change management process. If people resist it, it doesn’t matter how much you spend on consulting or how good the PowerPoint is. That money is wasted.

Jeremy: And in small business, the boss isn’t always in the room. They’ve delegated to someone who might feel threatened.

Ruby: That’s why we always require a project sponsor from the leadership team. Someone at the top has to want this — they’re signing off on it, they’re responsible for the ROI, and they’re responsible for communicating to the team why it’s happening.

When I was a GM internally, I could communicate that directly. As an external consultant, I need my equivalent inside the business. Someone who sees it the way I do. We do a lot of that alignment work in the feasibility study stage.

If I can tell someone’s only doing this because their boss told them to and they just want to tick the box — I won’t take it on. The failure rate is too high and it’s not fair to anyone.

Jeremy: And on the other side — do you use it to get rid of people, or to grow the capacity of the team?

Ruby: If somebody says they want to do this so they can fire a hundred people — that doesn’t sit well with me. What I hear from most clients is they want more capacity. They want to take on more work without burning out their team. Be transparent about that, and people get on board. It’s a very different conversation.

Closing Thoughts

Jeremy: Would you say your business is still in an early phase?

Ruby: The market is still working out what AI actually is versus the hype — yes. But in terms of our tech maturity, some of my team have been working with machine learning since 2016. Back then they didn’t call it AI. Now they do. And my implementation skills come from leading large projects for years. That part isn’t new to me.

The market confusion is real though. People don’t know whether to just buy a tool and hope for the best. A lot of education still needs to happen.

Jeremy: I think you need to be a bit less selective early on — I had the same issue when I started. If you’re too rigid about criteria, the people who don’t work with you might go with someone who doesn’t do it right, have a bad experience, and be done with AI forever. Sometimes you have to take on a bit more risk in the name of showing people what good looks like.

Ruby: Fair point. I probably am a bit too laid back. There was one client who came to me already having burned several hundred thousand dollars on someone who just kept throwing out buzzwords — agentic AI, Web3, blockchain — with nothing underneath it.

Jeremy: That’s exactly why street cred matters. What have you actually built? What have you touched with your own hands? The hype cycle in AI right now is the same as SEO was years ago. All noise, no proof.

Ruby: And that’s what I’d leave the audience with. A lot of small business owners are anxious about whether their business will exist in three to five years. We don’t know what the future holds, but what you can do right now is build your moat. Your data is your moat. Your processes are your moat.

Start building your data warehouse, your CRM, your systems. And when you’re ready — five to ten people — stop thinking about which tools to buy and start thinking about having your own AI, trained on your own data.

Jeremy: People too. I’d add people to the moat. I’ve got a team of ten and I can’t see AI removing them — I want to upskill them so they’re fully in the game.

Ruby: Exactly.

Jeremy: Well Ruby, I really appreciate you coming on the show. Tell people where they can find you one more time.

Ruby: People can find me at rubystrategic.co. I’m also active on LinkedIn. And yes — start with a free 15-minute call to see if it’s a good fit, and from there we work through it together.

Jeremy: Nailed it. We’ll check back in six months and see what you’ve been building.

Ruby: Thank you, Jeremy.

Key Takeaways

• AI needs a foundation before it’s useful. Data, systems, and people have to come first. Jumping straight to AI tools without clean data or operational basics is like building on sand. Ruby won’t take on a project until those foundations exist.
• Your data is your moat. Custom AI solutions trained on your own data are a business asset. Third-party tools send your data to someone else’s warehouse. Owning your data gives you something competitors can’t easily replicate.
• Change management is the hardest part. The tech is the easy bit. Getting the right internal champion, getting buy-in from the team, and framing AI as a capacity builder rather than a headcount cutter — that’s where most implementations succeed or fail.
• The translation layer is where the real value is. Most businesses have business people who can’t talk to tech people, and tech people who don’t understand the business. The consultant who sits in between — who can go from business problem to technical solution without losing either side — is the one who actually gets things done.

---