What We Found in Our First 10 AI Audits

In Episode 2 of the TriCities.ai podcast, Joe McKenna breaks down what his team discovered after walking into ten real businesses across the Tri-Cities and running AI readiness audits. The results were eye-opening. Every single business had staff using ChatGPT with no policy in place. Every one had manual processes bleeding hours each week. And not a single one was tracking what that lost time actually cost them. This episode tells the stories, names the fixes, and shares what genuinely surprised us.

Welcome Back

Hey, welcome back to the TriCities.ai podcast. I'm Joe McKenna. If you caught Episode 1, you heard us lay out the big picture. What AI actually means for small and mid-size businesses in this region. Today we're getting specific. Really specific.

Over the last couple months, our team has walked into ten businesses across Johnson City, Kingsport, and Bristol and done full AI readiness audits. We sat down with owners. We talked to front-desk staff. We watched how they work. We asked uncomfortable questions about what tools they're using and whether anybody approved those tools.

And I want to tell you what we found. Because the patterns were so consistent it was almost eerie. Like somebody copied the same problems and pasted them into completely different industries.

Before we get into it, I want to be clear. Every example I'm going to share today is real, but anonymized. I'm not naming businesses. They trusted us with honest answers and I'm going to respect that. You'll recognize the situations, though. Guaranteed.

And if you're listening to this thinking, 'I wonder what an audit would find at my place,' we do these for free. No catch, no sales pitch during the audit itself. Just a straight assessment. You can book one at tricities.ai/audit. I'll remind you at the end too.

All right. Let's get into it.

Three Problems. Every Single Time.

So after ten audits across different industries, different sizes, different cities, three problems showed up every time. Not most of the time. Every time. Ten for ten.

Number one: no AI usage policy. Number two: manual processes that AI could handle today. Number three: no measurement. No idea how much time and money they're losing.

Let me take these one at a time.

Problem One: No AI Policy

This one floored me. Not because businesses don't have policies. I expected that. What floored me is that staff are already using AI tools every day, and nobody's talking about it.

In every single audit, we found employees using ChatGPT. Sometimes for drafting emails. Sometimes for summarizing documents. Sometimes for things that made me genuinely nervous. And in every case, the owner either didn't know or had a vague sense of it but figured it was fine.

It's not fine. And I don't mean that in a scolding way. I mean that when you have no policy, you have no guardrails. You don't know what data is going where. You don't know if somebody's pasting client information into a free tool that stores everything. You don't know if your business is one careless copy-paste away from a compliance violation.

The thing is, the employees aren't doing anything wrong. They found a tool that helps them work faster and they're using it. That's what good employees do. The gap isn't at the employee level. It's at the policy level. And it's a gap that's shockingly easy to close.

Problem Two: Manual Processes Everywhere

The second pattern was manual work that should have been automated years ago. But I want to be fair about this. Five years ago, automating most of this stuff required expensive custom software or a full-time developer. That's not the world we live in anymore.

Today, AI can handle scheduling. It can handle intake forms. It can do follow-up emails. It can answer the phone. Not in a clunky, press-one-for-billing way. In a way where it actually understands what the caller needs and routes them or books them or answers their question.

In every audit, we found at least one process, usually three or four, where somebody was spending ten to twenty hours a week on tasks that an AI system could handle. And these weren't obscure back-office things. These were the core bottlenecks. The thing the owner complained about most was almost always something AI could fix.

Problem Three: Nobody's Measuring

This was the quiet one. The problem nobody thinks about because they're too busy dealing with the other two.

Not a single business we audited could tell us how many hours per week they spent on a specific manual process. Not even a rough number. It was always, 'A lot.' Or, 'Too many.' Or my personal favorite, 'I don't even want to think about it.'

And I get it. When you're running a business, you're not sitting there with a stopwatch. But here's why it matters. If you don't know the number, you can't calculate the cost. And if you can't calculate the cost, you can't make a smart decision about whether a solution is worth the investment.

So part of every audit is helping businesses actually put numbers to this. Real numbers. Hours, dollars, opportunity cost. And when people see those numbers for the first time, the conversation changes completely.

The Stories: What We Actually Found

Okay. Let me tell you some stories. These are the real audits, anonymized but accurate. I want you to hear what it's actually like when we walk through someone's front door and start asking questions.

The Dental Practice in Johnson City

So we walk into this dental practice in Johnson City. Nice place. Good reputation. Busy. That's actually the problem. They're too busy.

They've got two front-desk staff, and both of them are drowning. The phone rings constantly. Scheduling, rescheduling, confirming appointments, answering insurance questions. We tracked it. More than fifteen hours a week, just on scheduling calls. Between two people. And that's not counting the time they spend on patient communications, follow-ups, and intake paperwork.

Here's where it gets interesting. We asked about AI tools, and one of the front-desk staff kind of got quiet. Then she said, 'I use ChatGPT sometimes to help me write emails to patients.' She looked nervous saying it, like she might be in trouble.

She wasn't in trouble. But she should have been worried. Because she was pasting patient information into ChatGPT. Names. Appointment details. Insurance information. All going into a free tool with no HIPAA compliance, no business associate agreement, no data protection whatsoever.

She had no idea. And honestly, why would she? Nobody told her not to. Nobody gave her guidelines. She just found a tool that made her job easier and used it. Which, again, is what good employees do when they're overwhelmed.

She found a tool that made her job easier and used it. That's what good employees do when they're overwhelmed. The failure isn't hers. It's the absence of a policy.Joe McKenna

So what did we recommend? Two things. First, a HIPAA-compliant AI usage policy. Clear rules about what tools are approved, what data can go where, and what's off-limits. That's a one-time engagement. Second, an AI receptionist system. Something purpose-built for healthcare that can handle scheduling, confirmations, rescheduling, and basic patient questions. All HIPAA-compliant. All integrated with their existing practice management software.

Cost? The AI receptionist runs between thirty-five hundred and seventy-five hundred dollars to set up, depending on complexity, plus two-ninety-nine to seven-ninety-nine a month for ongoing service. The policy is a separate engagement.

The ROI? That's the part that got the dentist's attention. They were actively looking to hire a third front-desk person. That's at least thirty thousand a year in salary, plus benefits, plus training time, plus the fact that you can't hire someone who works nights and weekends handling online scheduling requests.

The AI receptionist costs a fraction of that hire. And it doesn't call in sick.

The HVAC Company in Bristol

Next up. HVAC company in Bristol. This one's a classic Tri-Cities story. Good company. Family owned. Reputation built on solid work. But they've got a problem that's costing them jobs.

When a lead comes in, whether it's a form on their website or a phone call, the average response time was four hours. Four hours. And honestly, sometimes longer. Because here's the deal. The owner's usually on a job. The techs are on jobs. There's one office person and she's juggling dispatch, invoicing, and parts ordering.

In HVAC, if somebody's AC goes out in July, they're not waiting four hours. They're calling three companies and going with whoever picks up first. So this business was losing leads to slower competitors who just happened to have someone near a phone.

Now, the owner had actually tried to fix this on his own. He'd bought a chatbot for his website. Paid for it, set it up, got it mostly working. But he couldn't get it to integrate with his scheduling system. The chatbot would collect lead info but then it just sat there. Somebody still had to manually transfer it. He used it for about two months and gave up.

That story came up a lot in our audits, by the way. I'll talk more about it later. But the short version is: a lot of business owners have tried AI tools on their own, gotten sixty percent of the way there, and then hit a wall because the last forty percent is integration and customization. And that's where they need help.

What we recommended: a lead capture bot that actually connects to his scheduling and dispatch system, plus automated follow-up sequences. When a lead comes in, the bot responds within sixty seconds. It asks qualifying questions. It can book an estimate on the spot. And it follows up automatically if the lead goes cold.

Setup runs two to five thousand dollars. Monthly service, one-ninety-nine to four-ninety-nine. The ROI on this one is almost unfair. He was losing an estimated three to five jobs a month to slow response times. At his average ticket size, that's tens of thousands in lost revenue. The system pays for itself the first month.

And here's the thing he said that stuck with me. He said, 'I knew I needed something like this. I just couldn't get it to work.' That's not a technology problem. That's a support problem. And that's exactly why managed AI services exist.

The Law Firm in Kingsport

This one made me the most nervous. A law firm in Kingsport. And I'm going to tell you why it made me nervous upfront.

When we did our audit, we discovered that staff, including paralegals, were copying and pasting client documents into ChatGPT. Full documents. Case details. Client communications. Attorney-client privileged material. Going straight into a free AI tool.

Let me be very clear about what that means. Attorney-client privilege is one of the most fundamental protections in law. If confidential client information goes into a third-party system without proper safeguards, you've got a potential ethics violation. You've got a potential malpractice issue. You've got a very bad day.

And again, the people doing this weren't being reckless. They were trying to be efficient. A paralegal told us she used ChatGPT to help summarize long case files. 'It saves me an hour on each one,' she said. And she was right. It probably did. But the risk was enormous and nobody had ever flagged it.

The other problem we found was search. Their own files. This firm had years of case files, briefs, memos, and precedent research scattered across shared drives and filing systems. When someone needed to find a relevant precedent from a past case, it took twenty minutes on a good day. Sometimes longer. And sometimes they just gave up and re-did the research from scratch.

We recommended two things. First, an AI usage policy. Specific to legal. What tools are approved. What data can go where. What constitutes a violation. What the consequences are. That's a seven-fifty engagement. Not expensive when you consider what a single ethics complaint costs.

Second, an internal knowledge assistant. Think of it like a private, secure ChatGPT that only knows about your firm's files. It's trained on your case history. Your memos. Your research. And it stays inside your systems. Nothing leaves. A paralegal can ask it, 'Find me precedents related to wrongful termination in the manufacturing sector,' and it pulls from your own library in seconds.

Setup for the knowledge assistant is around twenty-five hundred. Monthly service runs one-forty-nine to four-ninety-nine depending on the size of the document library and number of users.

Now let's talk ROI. Attorneys at this firm bill between a hundred fifty and four hundred dollars an hour. If we save five hours a week across the firm, which is conservative, that's three to eight thousand dollars a month in recovered billable time. Not theoretical savings. Actual billable hours that are currently being burned on document searches and manual summarizing.

At $150-$400 an hour, five hours a week saved across the firm is $3,000-$8,000 a month in recovered billable time. The knowledge assistant pays for itself before the first invoice goes out.Joe McKenna

One more thing about this audit. The paralegals were the most enthusiastic people in the room. Not the attorneys. The paralegals. They immediately understood what a knowledge assistant could do for them. They were the ones asking, 'When can we start?' Because they're the ones doing twenty minutes of searching per query, multiple times a day. They felt the pain most directly.

The CPA Practice

Last story. A CPA practice. And if you've ever dealt with a CPA at tax time, you already know where this is going.

The biggest time sink wasn't taxes. It wasn't advisory work. It wasn't even bookkeeping. It was chasing documents. Every year, same story. Clients don't send their stuff. Or they send half their stuff. Or they send a shoebox of receipts and a prayer.

The firm's staff would send requests. Follow up. Follow up again. Play phone tag. Send another email. Get a partial response. Ask for the rest. Tax season was three months of chaos, and at least a third of the chaos was just getting clients to hand over the documents they needed.

And here's the thing that nobody talks about. A big part of the problem is that clients don't know what documents to send. They get a list from their CPA and it might as well be in a foreign language. 'Schedule K-1.' 'Form 1099-NEC.' 'Mortgage interest statement.' Normal people don't think in tax forms. So they freeze, or they send everything and hope for the best, or they ignore the email until March.

What we recommended was a client intake automation system. An AI intake agent that reaches out to clients, asks them questions in plain English, and figures out what documents they need to send based on their answers. Not 'Please provide your Form 1099-NEC.' Instead, 'Did you do any freelance or contract work this year? Did anyone pay you for a side job?' When the client says yes, the system says, 'Great, you should have gotten a form from them showing what they paid you. Can you upload that?'

It speaks human. And it follows up automatically. Politely. Persistently. Without anybody at the firm having to lift a finger.

This was the most expensive recommendation we made across all ten audits. Setup runs about seventy-five hundred. Monthly service is three-ninety-nine to fifteen hundred, depending on client count and complexity. But for a CPA firm that's spending weeks of staff time chasing documents, the math works out fast. And the reduction in tax-season stress alone is worth something that doesn't show up on a balance sheet.

What We Recommended Across the Board

Let me zoom out for a second. Across all ten audits, our recommendations fell into a few buckets.

First, every single business needed an AI usage policy. Every one. This isn't a nice-to-have. Your staff is already using AI. The question is whether they're using it safely or whether you're one clipboard paste away from a problem.

Second, most businesses had one or two processes that were screaming for automation. Scheduling. Lead capture. Client intake. Follow-ups. These aren't futuristic use cases. These are solved problems. The technology exists today, and it's affordable.

Third, everybody needed measurement. You can't manage what you don't measure. And you definitely can't justify an investment in AI if you can't put a dollar figure on the problem it's solving.

The cost ranges varied. On the low end, an AI usage policy is a few hundred dollars. On the high end, a full intake automation system might run seventy-five hundred to set up with a monthly service fee. But in every case, every single case, the ROI math was positive. Usually dramatically positive. We're not talking about breaking even in eighteen months. We're talking about paying for the investment in the first month or two.

What Surprised Us

Okay. Now let me tell you what surprised us. Because some of this genuinely caught me off guard.

First surprise: everybody's heard of ChatGPT. That wasn't the surprise. The surprise is that almost nobody had a policy about it. I expected maybe half the businesses to have some kind of guideline. Even a one-page thing. 'Don't put client data into free tools.' Something. But it was zero. Zero out of ten.

Second surprise: the businesses that were already using AI informally were the most interested in managed services. You'd think it would be the opposite. You'd think the ones who'd already dipped their toes in would say, 'We've got it handled.' But no. They were the ones who knew enough to know they were out of their depth. They'd seen what AI could do. They just couldn't get it to work reliably on their own.

The businesses already using AI informally were the most interested in managed services. They'd seen what it could do. They just couldn't get it to work reliably on their own.Joe McKenna

Third surprise, and this was the big one: the number one fear wasn't cost. I went into these audits expecting price resistance. 'That's too expensive.' 'We can't afford that.' But that almost never came up first. You know what came up first? 'What happens if it breaks and nobody can fix it?'

Think about that. These business owners aren't hesitant of spending money on a solution. They're hesitant of spending money on a solution that goes sideways and then they're stuck. They've been burned before. By web developers who built a site and disappeared. By software vendors who didn't answer the phone. They don't want a tool. They want a partner.

And honestly? That's the entire managed AI pitch right there. You don't have to understand how it works. You don't have to maintain it. You don't have to fix it when something changes. That's our job. You just use it.

Fourth surprise: several businesses had tried AI tools on their own and gotten about sixty percent of the way there. That HVAC owner with his chatbot. A law firm that tried to set up an automated email system. A retail shop that experimented with AI-generated social media posts. They all got the tool working at a basic level. And then they hit integration. Or customization. Or both. And they gave up.

Sixty percent is a dangerous place to be, by the way. Because you've invested time and energy. You've seen the potential. But you can't get the last mile done because it requires technical skills you don't have and shouldn't need to have. That's frustrating. And it makes people cynical about the next person who walks in and says, 'AI can help.'

Fifth surprise, and this one was genuinely counterintuitive: smaller businesses, under ten employees, had more appetite for AI than the mid-size ones. We thought it would be the other way around. More resources, more openness. But no. The smaller shops moved faster because there's less bureaucracy. The owner makes the decision and it's done. At the mid-size places, there were committees. Approvals. 'Let me talk to my partner about it.' 'We'll discuss it at the next board meeting.'

The small businesses said, 'Can you start next week?' The mid-size businesses said, 'Can you send us a proposal?' Different energy entirely.

Why This Matters for the Tri-Cities

Let me step back and talk about why I think this matters beyond any individual business.

The Tri-Cities has always been a place where people work hard. That's not a cliche. It's just true. The businesses here are built on effort and relationships and showing up every day. And that's great. But right now there's this gap opening up between businesses that are figuring out how to use AI and businesses that aren't. And it's not a future thing. It's happening now.

That dental practice is competing against corporate chains that have AI scheduling baked in. That HVAC company is competing against franchises with call centers. That law firm is competing against bigger firms with research tools that cost six figures. The AI isn't optional anymore. It's the table stakes.

But here's what I believe. I believe that local businesses can compete. Not by outspending the big players. By being smart about where AI gives them the biggest advantage. A dental practice doesn't need a million-dollar system. It needs a thirty-five-hundred-dollar receptionist that never misses a call. An HVAC company doesn't need a call center. It needs a bot that responds in sixty seconds.

That's the whole point of what we're doing. Making the tools accessible. Making the pricing real. Making sure somebody's there when it needs fixing.

What a Free Audit Actually Looks Like

I want to take a minute and explain what actually happens when you book an audit with us. Because I think people hear 'free audit' and they imagine either a fifteen-minute phone call or a high-pressure sales visit. It's neither.

We come to your business. Or we do it over video if that's easier. We spend about ninety minutes. We talk to whoever runs the day-to-day operations. Not just the owner. The people who actually do the work. Because they know where the bottlenecks are.

We ask about your processes. Your tools. Your pain points. We look at where time is going. We ask about AI tools your staff might already be using. And then we go away and put together a report.

The report tells you three things. Here's where you're losing time. Here's what we'd recommend. Here's what it would cost and what you'd save. That's it. No obligation. No contract. If you want to move forward, great. If you want to take the report to someone else and have them implement it, also great. The information is yours.

We do this because we genuinely believe that once business owners see the numbers, the decision makes itself. We don't need to hard-sell anybody. The math does the talking.

What's Coming Next

In Episode 3, we're going to dig into one of these implementations start to finish. I want you to hear what it's actually like to go from 'we need this' to 'it's live and working.' The timeline. The bumps. The moment when somebody uses it for the first time and you see it click.

If you want to be part of that story, book your free audit at tricities.ai/audit. Seriously. No pitch. No pressure. Just a clear picture of where AI can help your business.

I'm Joe McKenna. This is the TriCities.ai podcast. Thanks for listening. Talk to you next time.

Sources

1. HIPAA compliance requirements for AI tools in healthcare settings Source
2. ABA guidance on attorney-client privilege and AI technology Source
3. Book a free AI readiness audit for your Tri-Cities business Source