[00:00] Rabia: Welcome back to another episode of Tech Unhinged, where tech gets human. Today I’m joined by Patrick Jean, Director of Sales Engineering at Crexendo. Patrick has spent over 15 years in sales engineering — from Comcast to fiber, and now Crexendo — helping enterprise companies adopt new technologies the right way. He’s one of those people who knows how to take big executive visions and turn them into solutions that actually work, which is exactly why I’m so excited to have you with us today, Patrick.
[00:29] Patrick: Thanks for having me, Rabia.
[00:31] Rabia: You’ve spent a big part of your career in technical sales leadership. What’s the most surprising or funny misunderstanding you’ve ever had with a client about AI?
[00:42] Patrick: The funniest thing is that a lot of people look at AI tools in a demo and extrapolate how they’d work in a practical environment. For example, they’ll see something like AI sentiment analysis in a demo, and customers think the sentiment analysis is a way for an agent to course-correct during the interaction — that it’s shaping the agent’s behavior in real time. Really, it’s meant to review the interaction afterward and coach the agent on what they did right or wrong. So the weirdest expectation I see is, “We thought this tool would make our agents change their behavior in real time.” That’s where I have to step in: AI tools are great, but they’re not magical — they don’t change people’s personalities. A lot of the noise feeds that misunderstanding, and it extends well beyond sentiment analysis. AI has a tendency to make people think a lot of things can change in ways that aren’t grounded in reality.
[02:17] Rabia: If we break down the jargon a bit — we hear terms like “AI governance” and “responsible adoption” tossed around a lot. What do you actually mean by those terms for enterprises?
[02:31] Patrick: When we’re talking about governance and accountability, this is where my department comes in. There are so many tools being added and promoted so fast that the compliance, laws, and regulations around them haven’t been fleshed out. So it’s on the providers to set the stage and keep things practical, as opposed to what a customer might wish for or be sold. My contribution is to step in — not to be a blockage or slow down progress, but to be “the grown-up in the room.” I ask the right questions: AI can do this, but where is it getting the data? Is it safe? Does the integration give it access to more data than it needs? How is that data contained? In a medical environment, for example, how does that fit with HIPAA compliance? Or PCI, depending on the industry? Things can look great in a demo — it’s on us from the technical side to understand the back end, what security features are established, and what compliance is maintained.
[04:12] Rabia: When executives say they want to build an “AI-first company,” what’s usually the simplest way to explain the gap between their ambition and the reality?
[04:25] Patrick: The first thing I analyze is the workflow — how does AI fit into it? AI doesn’t exist in a vacuum, especially with SaaS systems where you tack on different tools to make the whole flow work. So I ask: where does AI fit in that flow, and does introducing it disrupt the other tools already there? For example, say they want an AI receptionist to take incoming calls and route them. My question is: how does that AI integrate with the CRM that manages customer information? When a caller calls, their information usually gets presented to the agent. If we introduce a tool that can’t integrate with that system, then no matter how powerful it is, it’s not useful in any meaningful way. When an executive says, “I’ve seen this in a demo, I want it,” they’re not seeing what’s missing — because in a demo you have a contained environment with no integration needed. It’s my job to look at the bigger picture and the constraints we have to work within.
[06:31] Rabia: There’s a big debate in the tech landscape that the C-suite can be the biggest risk to AI adoption. From your perspective, how true is that?
[06:46] Patrick: It’s very true, and it varies by environment. If you have a C-suite made up of people who have the humility and curiosity to understand what they don’t know — and who bring dev members or engineers into those conversations — that’s the best case. But what I’ve seen in a lot of environments is executives who go to conferences, get captured by the AI boom, see competitors investing heavily, and say, “I want this tool right now, no matter what.” When a C-level executive sets strong requirements without consulting the dev team and just says “make it happen,” the dev team usually can — but it’s like asking for hot water in the shower and getting the pipes just warm enough. It’s a bandage on the problem. I’ve seen environments where the C-suite doesn’t really understand what it takes to build and maintain systems; they just want to see results. Executives with a technical background, or an open-door policy where the tech team’s criticism is genuinely heard, can avoid that — but that’s not what I’m seeing at large.
[08:48] Rabia: If you bring it down to a few options — where do executives most often underestimate the complexity of AI? Is it integration, compliance, or the reality of scaling?
[09:01] Patrick: The biggest one is integration — it’s the biggest topic next to AI itself, and it’s really technical, so it’s not part of the conversation. Those AI tools aren’t developed by the systems using them; a telecom or a CRM company is using and configuring them. When we sell a system to a university, a school, or a medical facility, they already have systems they need to keep using. So how does the new AI tool integrate and use information from those systems as part of the workflow? That’s not usually discussed from pre-sale until implementation. And integrating isn’t just creating a web hook — once the systems connect, how is the data managed? Is it clean? Because no matter how powerful the tool is, if you feed it bad data, you get bad results. The other part is the guardrails on security and compliance. Because AI tools are appearing so fast, the compliance around AI safety isn’t there yet — so it’s on the company providing the service to decide what guardrails and “gotchas” to be aware of. Those are the two areas people should focus on more.
[11:41] Rabia: Have you come across a situation where executive enthusiasm pushed a company into an AI initiative that was unrealistic or poorly planned? Can you share what happened?
[11:54] Patrick: Without naming names — we’re working right now with a major medical provider. They came off a system they’d purchased about a year ago. It was sold with a great demo; the system did what they wanted on screen. But once implemented, it turned out it didn’t really communicate with their EHR system — the system they use to manage patient information, with all the HIPAA compliance around it. Six months down the line, they found their process was worse than before, because the nurse or agent still had to rely on the old way of handling patient information and add new layers — email, web chat, calls. Instead of one portal that integrates everything, they now had multiple screens and systems, and the more systems you have, the slower the process. So now they’re very careful — they want proof of concept at every step, because the cost of a badly configured system in a major medical environment is unimaginable. It’s like having a teacher check your homework every step of the way; it makes us better as an implementation team, but you can tell they’ve been burned before. We have to show integration in practical environments, under stress, and what happens when the system breaks and how fast it recovers. We’re going through something similar with a school that was promised a web chat that couldn’t deliver — again, integration not following through and data not managed the right way.
[15:36] Rabia: Enterprise IT and AI spend is pushing $15 trillion or more. Do you think AI adoption is sometimes more about optics — the shiny-object syndrome — than real outcomes?
[15:53] Patrick: There’s definitely an aspect of that. There’s a lot of boasting — “my company is investing this much in AI” — but how much of that is actually practical? It’s murky, because you’re betting on things that haven’t had time to show a return, and there’s a lot of wishful thinking. But it’s necessary, because you’re betting on tools that could genuinely change how entire industries operate. So I think the spend is necessary — but the guardrails we talked about need to be in place. If a company has a billion-dollar AI budget, my advice wouldn’t be to reduce the investment; it would be to understand where it’s going and put guardrails around how it’s spent.
[17:08] Rabia: With 15 years of expertise in tech and sales, in your role as a sales engineer, how do you balance supporting executive vision with being honest about what’s realistic — and when do you know it’s time to push back?
[17:29] Patrick: I have a set of pointed questions I ask. I look at: what problem is the AI solving? Is there a simpler way to solve it without the complexity of AI and integration? Are you asking for this tool because it sounds good, because a competitor has it, or because you saw it in a demo — or does it solve an actual problem you have? Once I have that answer, I ask how the AI fits into the whole workflow from beginning to end. I try to understand the current environment first, then see where AI fits — rather than pushing a tool and hoping it works. So with a C-suite or a director, I’ll say, “Walk me through your process. Tell me what happens when I call your hospital. Who do I speak with? How is the call routed? How is my information captured — does the agent type it in, or is it automatic? Where does that data go? How are you keeping it safe? And when you run reports, do you actually need all that data, or are you capturing it for its own sake?” The longer that conversation goes, the less friction we get on the back end and the more tied to reality we are. If those questions aren’t being asked, you’re at the whim of the people selling you the tools.
[19:50] Rabia: When it comes to technical sales leadership, how do you handle AI adoption across the different teams involved?
[20:10] Patrick: Each group has a different part to play. The C-suite doesn’t need to be deeply technical, but they need to understand the requirements and what’s possible and what’s not. The sales engineering team educates the sales reps, who need to understand the practical application and limitations of the tools — they don’t need to know how the sausage is made, but they need to know what’s available and what those tools can and can’t do yet. If a tool is still in beta and hasn’t been stress-tested, they have to say so. Then the implementation team has to understand what the customer wants and relay it to the developers building and configuring the system. And the devs need to be very technical — they need to know the AI tools, the integrations between them and the various systems, and the parameters around data and data access. Not everyone will know everything, so we separate what each department needs to learn. When I run training, I have specific sessions for the C-suite, for the sales team, for the implementation team, and so on.
[22:58] Rabia: You work with a lot of international clients. How do you see regional regulations — like Europe’s AI Act — shaping executive decisions differently from the US?
[23:10] Patrick: Things will end up converging, because you can’t be as siloed as you were even a decade ago — companies are more and more global. There will always be some regional regulations, but on AI and how data is managed, you’ll see convergence, even if we haven’t yet. Internationally, some places are ahead and some are looser with regulation — and the places where AI is being pushed everywhere are usually where the issues surface first. Either extreme is a problem: if you restrict too much, you don’t get to stress-test tools in practical environments, because things work fine in a lab where you’re not bound by regulation — it’s real-world use where regulation steps in. I think the US market — the biggest one I work in — is roughly the sweet spot: there’s some regulation, we need better, but compared to the Asian market and others, the US has close to the right balance, even if it’s not quite where it needs to be.
[25:30] Rabia: On diversity and inclusion — AI and tech leadership are often criticized for lacking diversity. How does more inclusive leadership change the way organizations adopt AI responsibly?
[25:53] Patrick: Great observation. One of the best AI communicators I’ve seen was a speaker at a conference last year — she had a genuinely human way of explaining AI tools that I’d never seen before, and we need more of that. In a tech world dominated by male engineers, there’s a gatekeeping idea that to talk about AI you have to be very technical — which isn’t needed, and can even be a bad thing if you only see things through a technical lens. There should be room for more diversity and more ways of communicating AI, because people relate to those who can talk to them in terms they understand — and that’s what’s lacking from the C-level down to the development teams. Every time we bring in more women, more people of color, more different voices — as long as they understand and see the big picture — it helps. The best communicators understand the system but don’t try to prove how technical they are; they explain in real, day-to-day terms how the tools help. I agree with your assessment — I haven’t seen enough diverse representation in that space; it’s a continuation of the tech space.
[28:07] Rabia: Another blind spot: vendor lock-in versus flexibility. How many enterprises rush into AI vendor partnerships without thinking long-term? Do you see lock-in as a blind spot executives are missing?
[28:42] Patrick: Yes, that’s a big one. AI is changing so fast that a tool doing a function today may have a much better version out in six months — and swapping it isn’t like changing a tire. If your system is hard to integrate, getting that new tool to work took a lot of back-end patchwork, and now you’re locked into a contract with no easy way to remove the old tool and add the new one. I’ve even seen environments where they keep the old tool and add the new one on top — so you have an older AI auto-attendant routing calls, plus a newer AI agent taking and processing calls, layered on top. Being locked in creates all kinds of back-end issues that make the system look bad under the hood. The best systems are built for integration — you can take things out and put things back in, and update from version one to version two easily. That’s what people should focus on. Which is why I always bring AI back to integration — integration is really what makes AI what it is, until we get to AGI.
[31:05] Rabia: You’ve built your career at the intersection of sales and tech. What’s one leadership lesson you’ve learned about bridging vision with execution?
[31:17] Patrick: I was in technical sales while going to school for engineering, so I was building my sales career and my technical knowledge at the same time — two very different mindsets that sometimes conflict. The biggest challenge is understanding tech but being able to talk to people who have no desire to know tech but know they need it. Over my career I’ve developed the skill to geek out with the devs and then translate that for executives or sales leaders who just want it to work — and that translation isn’t as simple as it seems. The biggest lesson: I’ve watched sales engineers and developers talk to executives above their heads, thinking they’re making sense because it makes sense in their own minds — but it doesn’t land, because you’re giving fine detail to someone who thinks in the big picture. Being able to parse out the detail matters, but there’s a very different way to communicate with non-technical people. I didn’t decide one day to learn that — over the years I realized that the more I know about the product, the more opportunities I can find, but I don’t want to talk over people’s heads. That’s the biggest lesson I’ve learned.
[34:29] Rabia: Fast-forward five years — what will we look back on as the biggest mistake enterprises made with AI adoption?
[34:47] Patrick: Five years — though with AI, we predict something for ten years and it happens in five; things always move faster than we anticipate. But looking back, the mistakes will be: one, not doing the prep work to get their data AI-ready. Two, not training their development and implementation teams on how to integrate the faster, smarter AI tools with their existing systems. Three, being locked into AI tools that weren’t built with integration in mind — bolted on top of systems you can’t remove. And underneath all of it: not listening to their sales engineers — the team in the middle between reality and wishful thinking. That’s where the mistakes will be made.
[36:08] Rabia: We’ve recorded you — a year from now we’ll redo this and tell the world how Patrick already predicted it.
[36:20] Patrick: How my prediction came to pass — and don’t shoot the messenger.
[36:30] Rabia: Last question. If you could give one bold piece of advice to today’s executives on how not to derail AI adoption, what would it be?
[36:43] Patrick: It’s simple to look at a tool — at a conference, in a commercial — and see the vision. My best advice to executives is to really listen to their internal team, the people implementing the tools. And to ask themselves a few questions. One: what specific problem am I trying to solve, and is this tool uniquely able to solve it — or is there a simpler way? Two: how does this tool fit into my current environment? When you acquire AI, you’re not acquiring a standalone system — it has to live within an ecosystem. So the best approach is, even before you get the AI, make sure your system is ready for integration: your data is clean, your process is laid out, you know what people need to do from point A to Z. Then bring in the AI — or partner with companies who will point these things out to you, not just sell you the tool. The ones who’ll say, “Let’s look at your system and your flow first; maybe we revisit this in six months, and in the meantime here’s what to do.” That would be my advice.
[39:01] Rabia: Patrick, this has been a valuable conversation. Thank you so much for your time and insights.
[39:06] Patrick: Thank you so much, Rabia. It’s been a pleasure.
