Proprietary Technology for Service Businesses: The Hidden Growth Lever

[00:00:00] Brad Friedman: Hey, Ghazenfer. Welcome to the Digital Slice podcast. How are you today?

[00:00:07] Ghazenfer Mansoor: I’m doing great. Thanks, Brad. Thanks for having me on the show.

[00:00:11] Brad Friedman: You bet. I’ve been looking forward to talking with you. I want to just jump right in because I don’t want to waste any time, waste any of your time especially.

[00:00:25] Talk to me off the bat about how the service-based business owners and other small business owners listening today can future-proof their business with AI automation and smart technology.

[00:00:45] Ghazenfer Mansoor: Well, that’s a big topic, so we can talk for a long time. I think service businesses are ideal for all of that innovation, because a lot is happening in the AI space.

[00:01:00] In fact, many services businesses are even way behind on technology as well. So it’s not about just using AI. AI is suddenly a hype as well for many people, or you can call it like a new, sexy word, and a lot of people just want AI as a tool, but I think it’s important to know what AI is, where AI can help my business, and how we take advantage.

[00:01:30] So there are many ways. Let’s look at how service businesses typically work. In many cases, as you’re talking about the service business growth, so you’re talking about improving the books. If you know the data, that’s the only time you know where my bottlenecks are, whether it is on the sales side, whether it’s the margin side, whether it’s customer service side, which customers are painful customers and which ones are good customers.

[00:02:00] So once you have that data, AI could be used to automate that. We can dig deep a little bit on that afterward. But the second part is improving the operations. That’s the second advice you typically get. So, writing processes, SOPs, document everything, whether it’s meeting notes or everything.

[00:02:27] And obviously on top of that, hire a coach or a consultant to help you do those. And there are a few other things. But the important part that many service businesses miss is the technology part. And interestingly, when we talk to people, one of the commonly asked questions that is provided is, “Oh, we have technology. We are using this CRM that is common for our industry.”

[00:02:53] And I asked this question, I said, “Is your competitor using the same software? Do they have access to the same tool?” They said, “Yes.” And then I said, “Okay, then where is the differentiator? There’s no difference. You’re both using the same playbook. So if you’re using the same playbook, how are you different from them?” Yes, you can bring some efficiency, some optimization in your process, but that’s not gonna still give you a differentiator.

[00:03:17] I’ve seen many of those stories where you have great service, you have good people, you have raving reviews, but you are still not getting ahead because there’s so much competition. There are 20 other businesses doing the same thing right in your ZIP code many times. So how do you differentiate?

[00:03:42] That’s where I believe that proprietary technology comes in. The proprietary technology is the single most important differentiator that can help service business in getting crazy growth as well as increased valuation.

[00:04:14] Because having proprietary tech, obviously it’s a big topic, but it is a different take, because once you have some asset, that’s when it makes a difference for your business.

[00:04:30] Brad Friedman: So give me an example of proprietary technology.

[00:04:38] Ghazenfer Mansoor: So proprietary could be in your business. You look at your workflow. Let’s say you are, for example, whether it’s a lawn mowing business, whether it’s a wealth management, there are certain things  let’s say you use a CRM to track your customers, but then you have another tool that may be doing something else.

[00:05:04] So your proprietary tool could be, for example, a tool that may integrate all of those tools and have a dashboard where you are communicating with your customers. I’m not saying there are dashboards available. You may not need it, but just as an example, assuming that’s not there. So you have to identify what that specific is.

[00:05:26] So it’s not always going to be a known item. That’s the key part, because if everybody knows what that secret sauce is, everybody can build it. Most people will just go for those regular tools that are available. So you do use those tools that are very specialized. I wouldn’t tell people to replace HubSpot because that’s solving your marketing problem.

[00:05:54] But how the data comes into your business, starting from how the customer interacts with your business, whether it’s filling a form to fulfill your request, or as it changes hands from one person in a company to another  all these things are critical. So where do you see there are bottlenecks?

[00:06:22] Where is the manual work happening? Are you still getting the leads in a traditional way, getting the forms filled, or is the customer fulfilling all of those needs maybe coming to you when it’s very clear? Are you reducing the friction of communicating with the customer, giving them feedback, getting feedback, all of those things?

[00:06:44] Depending on the specifics, there are so many things. How do you reduce human involvement in that? That doesn’t mean you replace the human, but how do you empower those people to be more efficient in doing the work?

[00:07:00] Brad Friedman: And can we use AI to help us find these bottlenecks and different areas where we could improve?

[00:07:11] Ghazenfer Mansoor: Absolutely. AI definitely will give you something based on what you tell AI. AI is relying on your data as well. It will give you recommendations based on that. But you’ll be surprised  if you start giving that data, you could use AI to create that strategy. You can ask AI to come up with those ideas based off of the data that you are providing, and you’ll be amazed.

[00:07:39] Yes, sometime it will be way over. But just like everything we talk about AI, AI is not about just generating something based on a request. So you gradually need to improve that. You have to provide more context, you have to provide more data, you have to have a right prompting that will give you.

[00:08:00] So even let’s say, if you have to write a blog, sometimes it could take you one to two hours even with AI. You could generate in the first five minutes by giving a topic, but that doesn’t mean it’s gonna give you. It needs to know how you create the cluster, which pages to connect for your service pages, for your portfolio, for your other blog.

[00:08:27] How do you create all of that? You need to provide all that content, all the context. With time it will be better. But initially, once you give all of those and you have a specific way of how you write, you have a specific way of maybe adding CTAs in a certain cases, what angle should it be?

[00:08:46] If you’re talking about, let’s say in our business, if we are writing something and it says the angle should be a healthcare, or we do a lot of work  fix broken projects where a good amount of work comes to us where customers come to us, they said, “We’re working with this developer”, or, “We have been working on this project for one year, it’s not going anywhere.”

[00:09:07] Or, “We have this user experience problem.” So many of those different problems. So we get a good amount of work on that space. So you could use AI to optimize some of that. So what we have to do is we have to tell those angles — for this stuff, we need to provide this other angle so that AI, when you’re writing those content, you have the right outcome.

[00:09:34] That’s when you really have the right output.

[00:09:39] Brad Friedman: Oh, that makes good sense. I had a call yesterday from the CEO of a wealth management business asking me about whether I could help them increase their valuation to help for a sort of a profitable exit strategy. And it sounds like some of the things that you’re talking about would help in that kind of situation.

[00:10:08] Are there other ideas that you have for things that we can do to prepare a service business for a profitable exit?

[00:10:16] Ghazenfer Mansoor: Yeah, absolutely. Those are the right people. I think we should talk to people who want to sell in the next two to three years. We have seen many examples, many cases. For example, there’s one home care business that typically has a three times EBITDA valuation without any technology, without anything.

[00:10:38] Now, if you have another business, similar business, same size, same revenue, but now if that company has a tech, it is built for scale, and that would increase the value way more than just that. So we tell people, if they want to sell, yes, there’ll be an investment, but then that investment will help you get to the valuation that you want.

[00:11:03] Brad Friedman: Oh, that makes good sense.

[00:11:05] Ghazenfer Mansoor: And in our case, one of the common questions we get asked is, “Okay, you’re talking about not using these tools, or using these tools but creating something proprietary. How do we even know what we create? And how do we even start? We don’t even know.”

[00:11:28] That’s a common question I get asked all the time. There is no one-size-fit-all. It’s not picking another tool, it’s finding something unique for your business. Every business has a different DNA. So just picking anything generic would not work.

[00:11:57] So what we have to do is look at the processes and workflows and come up with very specific scenarios that identify those areas where efficiency could be brought in. So we have this process, we call it a blueprint process, where we review those flows and we come up with a strategy map that tells us, okay, here are the recommendations.

[00:12:26] Those are the things that could be done that could improve your business. And then, based on that, you can decide what to do. And it does not have to be spending a lot of money on day one to build a huge product. It’s a gradual process. You start with something very small and then gradually go towards it.

[00:12:51] Brad Friedman: Interesting. So in that blueprint process that you have, is that looking for ways that the business can automate?

[00:13:03] Ghazenfer Mansoor: Yes. The blueprint process would help businesses with  we provide that strategy for the business by looking at your processes and workflow. Here are the areas that can potentially be improved, and that’s where AI can make a difference, that’s where creating tech could make a difference, and that’s where the integration could make a difference.

[00:13:28] Every scenario is different. In some cases it may be just the integration. It may be, in some cases, a combination of all of those.

[00:13:37] Brad Friedman: And all of those things would help also differentiate that business from the business down the street that has the same valuation?

[00:13:49] Ghazenfer Mansoor: Absolutely. Absolutely.

[00:13:52] Brad Friedman: Okay. All right, so I understand that you do a lot of work in the healthcare arena. Are healthcare organizations looking at different things than, say, a financial services business? What does a healthcare organization do to prepare for AI?

[00:14:11] Ghazenfer Mansoor: Everybody is looking to get into AI, but the healthcare industry has been very cautious.

[00:14:22] As always. It’s not just now. Let’s say you talk to any doctors, any hospital, the very first thing comes in, “Oh, HIPAA.” They would not want to touch any application that is not HIPAA compliant, even if it does not have PHI data. So that’s a natural instinct where they are involved so much in HIPAA, so they want to see everything HIPAA compliant.

[00:14:46] So that’s natural. I think finances are somewhat similar. I don’t see much difference because obviously when you are touching financial data, it has to be PCI compliant. There are certain services, certain businesses that are taking some part of it, but then you still need that.

[00:15:09] But overall, the challenges are similar because you are touching people’s financial data. It could be your bank statement, it could be your mortgage and everything that you deal with. So that’s very sensitive data, and sometimes it’s more, I would say, more sensitive and more critical than even PHI data.

[00:15:32] So the similar challenges  they’re more cautious. I think the bigger part of the concern on the healthcare side, and I’m sure similarly on the financial side, is how do we make sure our data is not used by LLM to train? As you know, you load anything in ChatGPT or Claude or Gemini, wherever, they would use that data to train gradually.

[00:15:58] So people are adding that data left and right. Unfortunately, that’s all being used to train the data. And I think this is a good time for those businesses because they are getting the data automatically. But that’s a serious concern for the healthcare and financial businesses because you can’t…

[00:16:18] There are strategies to get away from that, and we can talk more about it another time if needed. But let’s say in healthcare, we talk about sometimes anonymizing it if it is that really critical data that needs to be trained. One of the common needs that come to us is, “Oh, we need a prediction for this thing.”

[00:16:47] One of the common demands we are getting more lately is, okay, we want to load the personal healthcare data and get more recommendations, whether it’s longevity data, whether it’s those concierge services as people are talking with the functional health data. So they want to see the recommendation based off all of that data coming in, and what can I do from this data, from a doctor’s side as well as from a patient’s side.

[00:17:21] So now that’s very sensitive personal data. Some of this data can be anonymized, so you don’t have to give the personal information. The first part is de-identifying that data. So as long as you can say, oh, this fifty-year-old Asian man has XYZ  with all of that information you can create a lot of good recommendations.

[00:17:51] In some cases, you don’t need to. There’s a concept, let’s say RAG, retrieval-augmented generation, where you could keep that data local, and you only use it to filter the results. So it’s a combination. You still use generative AI, but at the same time, you are not letting your personal data go onto the LLMs.

[00:18:18] And then obviously you can have your personal LLM deployed in your local environment. So there are many ways you could do it. Obviously, there are different cost consequences of each of those, but there are possibilities.

[00:18:33] Brad Friedman: But there aren’t any specific AI models that are HIPAA compliant?

[00:18:43] Ghazenfer Mansoor: I think they are, I would say.

[00:18:47] One of the requirements of HIPAA is signing a BAA. So when you sign a business associate agreement, you are committing to certain things. You are saying you will make sure the data is authorized, only people who have authorization are authenticated. There’s an audit trail of who accessed the data, what time, and encryption.

[00:19:18] So OpenAI does sign BAA, Gemini does sign, Google does sign. So when they are signing BAA for your business, then they are guaranteeing you that your data will not be used to train and will not be shared with anybody else. And that’s really a requirement  obviously, if somebody’s signing that, that’s my understanding that then you’re obligated to commit to that.

[00:19:47] Brad Friedman: Oh, okay.

[00:19:47] Ghazenfer Mansoor: So in that way, you can say yes, they do provide HIPAA compliance. But then there are all these different cloud providers, and you have to do some work. They also sign. I know Amazon, we work a lot with Amazon AWS, and they do sign BAA as well. But that’s more infrastructure as a service, not just a cloud.

[00:20:11] We still have to make sure that we secure the database, we secure the files, and how people are authenticating.

[00:20:21] Brad Friedman: Makes good sense. Okay, Ghazenfer, give me one or two takeaways you want listeners to take away from our discussion.

[00:20:31] Ghazenfer Mansoor: I would say if you’re in a service business and you are not building tech, you are building risk. You are competing, you’re just, I would say, floating and using the same playbook. So I would say definitely consider the tech in your business.

[00:21:18] Another one  on the AI part, you should automate wherever you can. Agents are a lot more popular nowadays, so you can use AI agents or create agents to optimize a lot of your business processes. And again, those are very easy to do. Those are not big undertakings. Those are small ones.

But on that, I do hear there are some people who have those concerns, where we see the reluctance from people in terms of AI. “Oh, is AI gonna replace our people?” That’s the first thing that comes in when something new comes in  a lot of people are worried about their jobs.

[00:21:46] The way we look at it, we say, “Well, it will change, but AI won’t replace your team,” it would empower that. So the way I look at it, I say, “Well, if you’re not using AI, then you may stay behind.” And the people who use it effectively will definitely outperform everybody else.

[00:22:10] Brad Friedman: Yeah.

[00:22:10] Ghazenfer Mansoor: So I would say use that as leverage. Use that to empower what you are doing. Increase your productivity, not consider that as a threat.

[00:22:21] Brad Friedman: Oh, that’s great. Okay. Tell us a little bit about your company and what you do and who you do it for.

[00:22:30] Ghazenfer Mansoor: Okay. So my company is Technology Rivers. We help companies build innovative software products. We do two things. One, helping businesses in process automation. We help them improve their businesses through AI and technology. That’s what we talked a lot about earlier.

And secondly, we help startup founders, entrepreneurs, product owners in creating innovative software products SaaS, web, mobile. Most of our work is in the healthcare industry where we assist health tech companies in developing HIPAA compliant web and mobile products.

[00:23:10] Brad Friedman: Okay, great. Where’s the best place for people to find you online?

[00:23:16] Ghazenfer Mansoor: So my company website is technologyrivers.com, R-I-V-E-R-S. And my personal website is ghazenfer, my first name, ghazenfer.com.

[00:23:30] Brad Friedman: Beautiful. Thank you.

[00:23:31] Ghazenfer Mansoor: And I’m on LinkedIn, easily accessible. My name is somewhat unique with these spellings. If you search it, you’ll find me right away.

[00:23:40] Brad Friedman: Okay. All right, Ghazenfer, thank you so much. I appreciate you talking with us today.

[00:23:48] Ghazenfer Mansoor: No, thanks for having me on your show.

[00:23:50] Brad Friedman: You bet.