User Retention: Engineering Trust Beyond the Download

[00:00:00] Michael Vickers: Welcome back to another episode of Becoming Preferred, the podcast where we dive deep into the strategies that help you level up your game and stay relevant in an ever-shifting marketplace. I’m happy to introduce our guest today, a man who sits at the intersection of high-level software innovation and human-centric design.

Ghazenfer Mansoor is the CEO of Technology Rivers, a powerhouse firm known for building everything from HIPAA-compliant healthcare tech to cutting-edge AI solutions. But Ghazenfer doesn’t just build software. He understands the psychology of why we use it. He is the author of Beyond the Download, where he breaks down the alchemy of creating mobile apps that people actually love and share.

Beyond his technical expertise, he’s a fellow storyteller as the host of the Lessons from the Leap podcast, uncovering the raw failures and bold breakthroughs that define the entrepreneurial journey. Whether you’re looking to scale your startup, automate your processes, or simply build a brand that resonates, you’re going to wanna take notes.

Join me for my conversation with Ghazenfer Mansoor Well, Hi Ghazenfer. Welcome to the program. We’re delighted to have you. 

[00:01:09] Ghazenfer Mansoor : Thanks Michael. Thanks for having me. 

[00:01:11] Michael Vickers: An interesting topic today, and we’re gonna talk about it because a lot of our listeners want to understand today’s technologies. They’re entrepreneurs, they’re business professionals.

They, how do I take my business to the next level? How do I grow my business in today’s competitive marketplace? And you’ve definitely got some experience there, but let’s go back in time here. You’re back in high school. You’re going to school, and you’ve got this path in front of you and you go, where am I gonna go?

How did Ghazenfer become Ghazenfer? How did you decide to go down this road and end up where you are? 

[00:01:45] Ghazenfer Mansoor : So I have a computer science background. I have a master’s degree in computer science, and I worked as an early engineer with two startups. So, I got involved in mobile in the very early stage around 2000 when it was just those flip phones.

So along the way when I worked with the startups, obviously that gives you more motivation towards working on your own ideas. So that gave me a push towards that. And every engineer likes to build things, and I started a recruitment software startup, and that’s how I got into entrepreneurship. So after that, I started this business in 2015, helping businesses and the product development, building it the right way the first time.

[00:02:29] Michael Vickers: How have you seen the technologies change? Like I remember back when I took computer science, it was Fortran and COBOL. I hate to date myself. And we’d just come off of punch cards, just go. So I’ve seen that migration then C and C+ and all the developing. And then I remember there was a whole period of time where, hey, I’m gonna be a coder in my next life, and my– That’s the way to go.

That’s the future. And that seemed to serve a purpose for a while. But now we’ve got AI and AI tools coming. How do you see the landscape? How has it evolved just in your time, you know, as a professional working in when it comes to tech? 

[00:03:02] Ghazenfer Mansoor : Yeah, I’ve seen the COBOL and Visual Basic time. I can definitely resonate with that.

Technology’s changing with amazing speed. I think- It’s one of those cases where it never stopped to amaze us. But this AI change is the biggest of all. So when I started my career earlier, it was an early web time, and then obviously everybody even building a website was a big thing. And, and then after that, mobile became a pretty big thing, right?

And now AI is changing everything. So definitely it’s a bigger change, but I, I would say it’s much bigger than everything else, that’s a big disruption in the industry, ’cause every other thing didn’t create as scariness as this one. So even though there are a lot more opportunities, you do see that in the general public you may hear a lot of that fear as well that AI is taking away all those jobs.

That was not the case before. Previously it was okay. Now all these opportunities are coming in, it’s gonna create more jobs. 

[00:04:03] Michael Vickers: Yeah. 

[00:04:03] Ghazenfer Mansoor : So. 

[00:04:04] Michael Vickers: Yeah. It always has. You call it a disruption. I think it’s the fourth great disruption that we’ve had, the industrial age, the technological age, the where we live today. And you’re right, I think people who don’t embrace the technologies will always lose to someone who does embrace them, but it’s just changing and leveling up our work.

It’s helping with that productivity in my opinion.  I remember websites, we needed coders to do websites, then we had the what you see is what you get. We had drag and drop, you could build your own website, and those came into play. And now even when it comes to coding, I’m not a coder, but I’ve had AI explain coding to me.

Explain it to me like I’m an eight-year-old, now a 12-year-old, now explain it like a 20-year-old, all within an hour, and it gives me a good sense of how this works or how to find this inside the code. So I think to your point, it can be disruptive if we don’t evolve, so we gotta keep evolving. And as business owners, that’s always the big challenge.

Where do we go and how do we do it? Now, in your latest book, Beyond the Download, you emphasize that getting, say, an app on a phone is only half the battle. For a business leader, what’s the single biggest reason users abandon a product, say, within the first 30 days? They look at it, they’re interested in it, but then they come off of it.

What are some of the issues and challenges that business owners face there? 

[00:05:17] Ghazenfer Mansoor : I think the basics are still the same. The foundational problem i- like, are you solving a customer’s pain? I always give examples, we both have seen the Craigslist time. Was it a good user experience? No. But how much traffic we had, like every time you go there, because they were solving one specific pain.

So when you solve pain, people will come back, and they will download, they’ll keep using. And some of those apps that we are using on a daily basis, they are solving some problems. They are providing some value to the users, and those users keep coming back. So yeah, that’s the gist of it, what problem you are solving.

And then the second part that changed after that Craigslist and during the Apple time is, is the user experience and design. In the past, the importance was user experience and simplicity was not as important, you just solved some problem. Now, you don’t have to just solve, but you also solve it in a really simple way.

So if the user has to think what to do on this app, it’s late. You’re not gonna get them back again. So your app has to be remarkable. It has to solve a problem, it has to provide value, it has to have an amazing user experience and design. It’s not the time when you have to watch the YouTube tutorial To learn how to use an app, which surprisingly a lot of people are still doing it 

[00:06:36] Michael Vickers: You ha- you focus on that user success, and a lot of companies for years it was all just about user acquisition.

How many acquisitions, that’s where the value came from. But then your churn rates were so significant ’cause people weren’t implementing it or internalizing it. So the abandonment usually happens because of that time to value. If they don’t experience the win within the first few minutes or an easy use, they’re moving.

They’re moving on to something that’s even easier. Which brings us to where we are today for how do we create strategies using today’s technologies of AI, for instance? How does AI strategy work for non-tech firms? So many of our listeners run service-based businesses. How should a non-technical entrepreneur decide between, say, buying an off-the-shelf AI tool or investing in a custom AI-powered solution?

[00:07:22] Ghazenfer Mansoor : So it’s a tricky situation.  I would say not every company has the same DNA. Every company is different. You have to look at a specific use case, specific problem you are solving. There are some, I would say, agents, some solutions that are customized. Those are commoditized. It does make sense to license them, use them.

You don’t want to be building everything. But at the same time, that’s also the differentiator in every service business. So if you are using the same system that your competition is using, that means you’re not really… How are you competing? Are you competing with people? Are you competing on pricing? Are you competing on speed?

Are you competing for better customer service? And all those could easily be replicated by your competitor and they can beat you, especially the one with more money. So how do you differentiate? That’s where technology can come in. So technology can make a big difference and help you 10X your business, not only bring efficiency, but also increase the value of the business as you’re going to sell in the long run.

It’s gonna make a huge difference. So when I say technology, what does that mean? It’s a proprietary technology we’re talking about. But at the same time, that doesn’t mean you replace your regular scheduling or regular routine software with custom. When we say technology, that means building a very specific workflow that is unique to your business.

It could be integrations with existing systems, whether scheduling, whether it’s CRM, CMS, whatever are the different tools you’re using, maybe integration with your payrolls and all of those things. And you need to have a dashboard, you need to have all those things then that help you move the needle in your business.

So your competitor may be able to copy the other tools that people are using, but they won’t be able to copy the unique workflows that you have. So we talk about when in the service business growth, you wanna look at what are the bottlenecks in your business? What are the manual workflows? Some of those could be replaced by existing tools, some may need to be created, some need integration, and every business has unique ways of doing things.

If you can optimize that, that would make a difference because now you can do things much faster than your competition. Now your team can identify the problem, you can identify the risk instead of learning about losing a customer if you get the red flags ahead of time because you have a system that monitors.

So there’s no one case I can specify, but it’s just unique. You have to look at your specific workflow processes and then come up with a strategy to implement those. Yeah. But it could be, again, it could be starting out with just one agent and then maybe adding into multiple agents or workflows. 

[00:10:04] Michael Vickers: Well, and that’s good advice.

I think, you know, it’s the old 80/20 rule. I think if an off-the-shelf tool solves 80% of your problem, great. There’s some good project managers, there’s some good programs. You can look at the workflows. If your competitive advantage, the thing that makes you preferred, lies in that remaining 20%, then you gotta build your own IP and build your own system.

And with the agents now, that seems to be also making that a little easier. We had bots for a while, and bots are still prevalent. We see a lot of bot, but people are getting bot fatigue, I think. They see a bot, the bot. Where an agent is kind of like a really advanced bot really, but we just call it something else, I think.

Let’s talk about trust level. One of the areas that you specialize in is HIPAA-level trust standards. So you specialize in HIPAA-compliant healthcare tech. As for those that aren’t in medical fields, what can entrepreneurs learn from healthcare about building digital trust and data security?

[00:10:58] Ghazenfer Mansoor : So any application that has a PHI, personal health information, it has to follow certain HIPAA regulations. So that includes obviously the physical controls as well as the software part. So we’ll only talk about the software part. So how do you exchange your data? Who accesses the data? Are the users authorized to access the data?

What time they access so that later on you could track the history and all of the audit logging? There are a whole lot of different rules that you have to follow when it comes to HIPAA. And in reality, even though the government regulations are on that healthcare data, that really applies to many of those businesses.

My business has personal sensitive data. Yes, it’s not PHI data, but all of the employees’ information is also as critical as my healthcare data. In fact, that is even more critical. So it does require compliance. So all the data– So I think as you’re via building these applications, we build that process and- structures of how to build those, applying those in a similar other businesses, even though they may not necessarily be officially HIPAA compliant, but making it compliant makes it much easier and build the trust and give the confidence.

[00:12:10] Michael Vickers: Yeah. I think security ought to be a feature, not a footnote. You know, not a, “Oh yeah, by the way, we’re also this.” Hey, by the way, we’re focused on your security, and feature it, highlight it, amplify it. It gives the user,  helps to build that trust. Hey, my data is secure. ‘Cause we trust our banks, we trust our systems.

We sometimes give up a lot of information. So I think having that HIPAA level security in our programs or apps I think will become critical. Let’s talk about scaling without breaking. Your company, Technology Rivers, you help startups scale faster. So when a business experiences a sudden growth spurt, where does technology come from, where does it usually break first, and how can we future-proof it?

[00:12:50] Ghazenfer Mansoor : So I think the foundation is the key. As we are building these applications, if we are just duct taping the systems, that’s where the problem comes because now the foundation was not set from day one. In today’s time with AI, with the cloud, all of that infrastructure is cheaper, it’s easier. Building scalable tech is much easier than it used to be.

So in fact, in our experiences, if you’re not thinking about the scalability from day one, you’re actually making it slower because that means you have that infrastructure, you have your pipelines and all of that infrastructure, you just automate a lot of those things, that you build it once and then you repeat that.

So a lot of those scalability things are much easier to implement because if you’re not doing it, then you are doing it more like a traditional way. You’re taking probably more time anyway. But that’s a big misconception we have seen when people are building these, okay, we just need an MVP and we don’t need any of this, but in reality you’re re-spending a lot of time.

Remember I talked earlier about when I got into this building, one of the goals was to build the product the first time the right way. One of the things we have noticed is that a lot of people coming from different backgrounds are building these products that are never finished because you build something and then keep updating, patching, duct taping, and you’re focused on fixing all those things for every new feature you’re building.

It’s taking much longer. So you’re running out of money just because your tech is taking so much time, obviously along with many other things. So the focus has to be building the product the right way the first time so that you can focus on growing your business rather than just fixing and patching the technology all the time.

[00:14:38] Michael Vickers: Right It’s very tough. I believe you teach that one in 10,000 become a true success. 95% of apps fail within the first year. So architecting, designing, how does an entrepreneur avoid that scope creep, if you will, where they have an idea and it takes forever and ever to get it on, and basically they run out of financial runway in order to get that app to market?

‘Cause sometimes it might be, “Hey, let’s get it out there and then keep evolving it and going through…” Is there the right approach? And how long on average should an app take? And I realize that based on the complexity it could vary, but what kinds of things are you seeing, you know, from concept to,  we’re rolling it out and we’re implementing it?What kind of timeframes should someone expect? 

[00:15:24] Ghazenfer Mansoor : The timeframe used to be different than now. So I’ll talk about the process first. So I think the most important part is really narrowing the scope of the MVP to building something that your customer needs. I think one of the bigger challenges, the feature race, is that people start building too many things and want to have a perfect product before you can launch.

So that perfect product, if you’re taking a year or two years to just build it and then you realize nobody’s using it, that’s a big risk. So we recommend you build a very tiny MVP, even if that means one feature or two features. One use case, two use cases that your customers really need out of the 50 features you want to build.

Start with one, put it out in front of the customer. It’s like you build the foundation, you have the sketch, and then you gradually scale. So one of the examples I gave, like if you’re building a house and your vision is a mansion, you want to start with a studio and then gradually expand. When I say studio, that means it’s a livable space that has a bathroom, kitchen, right?

It’s not a room that’s not usable. So make a usable MVP that may have just a login and two other things, or may not even have a login, but it’s doing something for your customers, and then you keep giving them features every week, every two weeks. And you’re exciting your customers as well because now you’re releasing every week, and they are seeing that something is happening.

Rather than you build something, nothing happened for six months, you lost trust in your customer because they don’t know when the next set is coming because you’re gonna make a big patch, and then there’s another whole learning, the data migration issues, all of those. So you want to focus on building just the tiny part.

That’s one thing. In the past, we always recommended the MVP should not be more than two to three months. Build it in two weeks, do testing, get feedback. Well, it could be a bigger one, but you can still even have a tiny version that you could test with your users. It could be an internal MVP, but it has to be live.

If it’s not deployed, it’s not live, your users cannot see it, even if it’s internal limited users, you can see some problems potentially down the road. Now with AI, even those things have improved. Now with white coding, now we can even do a lot faster. So one of the things we recommend, use one of these white coding tools, whether it’s Replit, Lovable, V0, there are many.

Create an MVP. Like rather than in the past, as you were clarifying the scope, you were creating those sketches, now use the white coding for initial proof of concept, and that could be done really quickly in hours and days, and then you gradually refine. Let’s say i- in a week to two week timeframe, you refine your flows in the white coding.

Now the next job is to take that and convert it into a real working MVP So that’s the flow we recommend now, building the initial MVP with the white coding in the first two to three weeks, and then expand it, make it production ready or make it HIPAA compliant and make it live in four to six weeks. But then that four to six week is much bigger than what it used to be, the three-month or six-month timeframe.

[00:18:43] Michael Vickers: For MVP, or which stands for a minimum viable product for those who aren’t focused on that, is there a danger of being too minimal? Like when does good enough actually hurt a professional brand? 

[00:18:53] Ghazenfer Mansoor : It really depends on a specific problem we are trying to solve. Let’s say if you are building a product that is– you think that you’re creating a replacement of whatever, let’s say a bigger tool, maybe a Salesforce or HubSpot or any of those.

Now, what is your selling point? So are you competing on, “Oh, I have five more features”? Even in that case, we say, “Well, you still have some differentiators. Why don’t you build something just on that differentiator?” Just like those that gradually, and maybe start with integrating with the existing one because change is difficult. It’s not easy for existing users of other products to just switch to your product. It has to solve something specific that others are not doing. What if you bring something that just integrates with others so that it’s complementary for them, and then gradually you add other features into those smaller products and expand?

But yeah, that’s really a risk as well because now, again, it also depends on your customers or the type of product. So you have to figure out who your customers are. Are they willing to buy what you are building? So if we cannot narrow those into a smaller one, that means our problem is complicated.

That doesn’t mean it won’t be solved easily, but,  you’re making a bigger problem. So anything that is bigger, obviously that becomes a bigger project, then that means a lot more stakeholders, a lot of other factors get involved. The smaller you break your problem into and solve it, then that small problem is much easier to solve.

[00:20:25] Michael Vickers: Are there different types of genres of– I want to use the word genre or types or categories of apps that have a higher chance of success versus those who don’t? You know, I’ve got tons of apps on my phones, and every once in a while I’m getting rid of some, and then I like to chase them if I see something new that’s focused.

I like those as well. But are you seeing somewhere, hey, we got too many of these, we don’t need these anymore, but this is open territory, this is a good field, this is a good area where an entrepreneur might want to add an application? 

[00:20:57] Ghazenfer Mansoor : So we don’t look at it from an app perspective. This is not really just an app problem. This is in general any business idea problem. 

[00:21:03] Michael Vickers: Right. 

[00:21:04] Ghazenfer Mansoor : So you have to identify the pain, what people are looking for. And if we know the examples of all these, let’s say, social networks and all of those. TikTok is solving one problem. Facebook is solving another one. They’re all different, and there are a lot of other competitive products as well.

Not everybody got the adoption. So it’s about what is missing in the industry, what is missing in that specific domain. And then to multiple categories, because there are some that are internal users. Let’s say you build an app that may work out really just for the specific types of user. Let’s say you’re a patient, so it’s just a necessity.

You have to use it, ’cause now you’re selling it to your whatever, hospitals or providers, versus those are consumer ones. Hard to scale, but once you get that virality, then it can go crazy as well. So it really depends on a specific problem you’re trying to solve, and what is your approach to solving that problem. Sometimes an app is just a complement of what you’re doing, versus an app is a necessity. Let’s say you talk about Uber, Lyft, those kinds. Without the app, obviously it’s a real-time experience. You’re tracking your ride and everything. You need an app. In some cases it’s a business, but an app is good to have.Some people will have it, some not, because they can still do it through the web or through other ways. 

[00:22:28] Michael Vickers: Is there– For entrepreneurs who are looking at perhaps adding an app or expanding and trying to scale their existing environment, is there a, “Hey, on the low side…” Like, how do we budget for those types of things?

Is there a good rule of thumb for budgeting? And I realize how long is a piece of string? How long do you want it to be? It’s like a marketing campaign or an ad campaign. And this is, I think, the thing that creates that blind spot for people, where they get nervous about it because they’ve heard the horror stories.

You know, millions versus tens of thousands. For someone to even consider, you know, if there’s a range and, hey, they should consider this on the low side, and then it can go anywhere. But what can usually solve a problem? ‘Cause it’s not like it used to be. Like, I, I don’t believe it takes millions and millions to do it, unless you’re building something massively complex. Are there any guidelines that we should be following? 

[00:23:16] Ghazenfer Mansoor : We have built apps that were, I mean, I think you, in today’s time you could build it maybe as low as 10,000 and m- maybe a few hundred thousand dollars. Again, it depends on what is included in it. Right. And sometimes the cost is not the first time.

The cost may be over time, and because you’re building it and growing. Because as we talk about, for example Uber, Lyft, and some of DoorDash, all of those, it’s not about just the app, it’s about the operations that run behind it, because now you’re servicing your customers as well. So building an app that is not in production is one price, but once it goes live, then you have to build the supporting software as well.

Maybe admin dashboard, maybe as the user is using, getting some kind of feedback. Now you’re tracking analytics. There are a whole lot of other things that you’re doing, so support that. So then your cost is more of managing than building the new features. So the starting MVP could be really cheap. 

[00:24:16] Michael Vickers: Yeah. Bring that. On your podcast, Lessons from the Leap, you talk about the bold decisions entrepreneurs make. What is one leap you see most business owners are currently afraid to take but probably should? 

[00:24:29] Ghazenfer Mansoor : Oh, that’s a hard one. Everybody is trying to get into the AI race, which is exciting, but at the same time some people are- here to take that leap.

But in our Lessons from the Leap, most people are the ones who are actually taking those leaps because those are the ones that we are talking. The one that are afraid are not much on the podcast yet. Yeah, I think from, I mean, if I talk about it from the Lessons from the Leap, it’s mostly the AI race that people are taking. It’s just a different approach people are taking. 

[00:25:00] Michael Vickers: Well, and I think you do talk about niche specialization as well. Many entrepreneurs fear that narrowing their focus kinda limits their market, but in reality, it can actually,  you avoid being a commodity when you start to focus and you start to niche down on it.

But let’s talk about AI, because I think that’s an important characteristic. I believe there’s the human side of AI. So as an expert in AI-powered solutions, where do you believe the human touch is most irreplaceable in the modern business journey? 

[00:25:27] Ghazenfer Mansoor : So AI is still obviously at, I would say, still at the early stage, even though it’s great, it’s in many cases perfect, but at the same time, it does need human involvement.

There’s a lot of hallucination, a lot of mistakes, so you do need human to verify those things before, especially on the critical cases. The way people are using AI, it’s also another big challenge, because if you’re not wording the context properly, if you’re not articulating your problem properly, you have to work on it to get to the result that you want. It takes time for that. So there’s a lot of training involved in that. Like, as people are using it, it’s learning and improving that as well. 

[00:26:11] Michael Vickers: And we’ve got two types. Like, I’m in agreement with you. I think AI is good for information. W- there’s generative and then there’s agentic. We’ve had agentic since, you know, 1967 was as far back as I can go where we’ve had agentic, where we get an agent to do it versus interacting with something like a ChatGPT, which we talk about.

I believe that humans, for instance, need them for empathy. We need them for interpretation. We need them for insights. We need them for accountability. AI can’t be accountable. So I think as a society, we have to evolve our EQ, all right, so that we can handle this high level of IQ that’s being generated for us and take care of it.

So it needs that balance, and I think that’s really what protects humans in the age of AI, and so we can thrive and survive by building our human skills. You know, if I’m talking to a nurse, if I’m getting a procedure, I want that nurse to hold my hand. And talk to me if I’m going through something that might be fearful.

If, you know, I was talking to a radiologist who now does– It used to take him four to five hours to diagnose charts, can now do four or five people within that same timeframe, where the AI can analyze at 99% accuracy. But I don’t want a computer explaining my chart. I want that radiologist talking to me.

So it seems like everything’s designed to level up our productivity, but we should really be focused on to stay competitive to stay distinctive and unique and enhance our human skills. Would you agree with that? [00:27:35] Ghazenfer Mansoor So AI is still obviously at, I would say, still at the early stage. Even though it’s great and in many cases very capable, it still needs human involvement. There’s a lot of hallucination and mistakes, so you need people to verify outputs, especially in critical situations. The way people use AI is also another challenge. If you don’t provide the right context or clearly articulate your problem, it takes time to get the results you want.

So there’s a lot of training involved. As people continue using AI, they’re also learning how to interact with it more effectively.

We see this clearly in healthcare, where most of our work is focused. Rather than spending hours analyzing information, AI can process the data first so doctors can spend more time communicating with patients and solving higher-value problems.

In several predictive health applications we’ve built, users can upload blood work and other health data to receive predictions. However, that creates regulatory risks. You can’t simply present predictions because someone has to be accountable for them. Instead, the system is designed around previous physician recommendations, and every result is reviewed by a doctor who can validate or modify the recommendations before they’re shared.

In some cases, those recommendations are converted into a more conversational format. One application even includes an AI-generated podcast where two virtual doctors discuss a patient’s report. After listening to the summary, the physician can focus on answering specific questions instead of spending an hour explaining every detail from scratch.

[00:29:42] Michael Vickers Interesting. If we’re building our own knowledge base or interacting with AI tools, many companies are concerned about data security. Organizations have policies about what information employees can upload because of privacy concerns.

How secure can we really feel when we’re entering personal information into an AI platform? Is that data protected, or should businesses still be concerned?

[00:30:12] Ghazenfer Mansoor Yeah, absolutely we should be concerned. You have to follow specific rules to make sure your data remains secure. In many cases, people simply upload everything into ChatGPT, and that information may be used for model training. If it’s your own personal data, that’s your responsibility. But when it comes to sensitive healthcare information or other confidential business data, you need a different architecture.

One approach is Retrieval-Augmented Generation (RAG). Instead of sending all of your internal documents to the AI, your sensitive information is tokenized and searched first. The AI only receives the limited context it needs to generate a response, rather than your entire dataset.

If you’ve uploaded hundreds of files into an AI system, you’ve probably noticed that it doesn’t search every document every time. Large language models have context limitations, which is one reason hallucinations occur. By limiting the information provided through RAG, you reduce hallucinations while also protecting sensitive data.

The next question is whether the AI provider uses your information for model training. That’s where additional agreements become important. We work with LLM vendors under Business Associate Agreements (BAAs) that include zero-retention policies, meaning the data isn’t stored or used to train future models.

By combining RAG architecture with these agreements, we know the data being transmitted isn’t retained or used for training. Another option is deploying on-premise LLMs that never send information to public AI services. Even then, RAG still provides benefits beyond security by improving accuracy and reducing hallucinations.

[00:32:36] Michael Vickers You’ve built a fence around it. This has been a fascinating conversation, Ghazenfer. We’ll include all of your contact information in the show notes, including ghazenfer.com and Technology Rivers. Listeners can also find your blog, podcast, LinkedIn profile, and your book, Beyond the Download: How to Build Mobile Apps That People Love, Use, and Share Every Day.

If you’re looking for practical strategies to build products that don’t just acquire users but keep them engaged, I encourage everyone to check it out. Thanks so much for joining us today.

[00:33:15] Ghazenfer Mansoor Oh, thanks for having me, Mike.