U.S. Manufacturing Today Podcast

Episode #82: Why Your Manufacturing ERP Isn’t Working: Connecting Systems, KPIs, and AI with Andy Janaitis

‍

On this episode of U.S. Manufacturing Today (powered by Veryable), host Matt Horine speaks with industrial engineer and Janaitis Business Systems founder Andy Janaitis about why implementing an ERP often fails to solve the underlying problems at smaller manufacturers. Andy explains how overly complex software, disconnected systems, and manual data entry can leave businesses with conflicting answers to basic questions like quote win rate. He discusses how connecting existing tools, defining consistent KPIs, and breaking forecasts into measurable inputs can help manufacturers identify bottlenecks and make better decisions before month-end. They also explore why AI tools need business-specific context and human feedback, and how businesses can reduce dependence on key employees without replacing their expertise. Andy closes with a practical 30-day starting point: identify the KPIs that drive the business and determine what's preventing them from being tracked reliably. He also shares information about a free 25-minute call and his webinar series.

‍

‍

‍

Links⁠

‍

‍

Timestamps

  • 00:00 Podcast Welcome
  • 00:41 Quote Win Rate Mystery
  • 01:43 Andy’s Engineering Path
  • 03:55 Why Manufacturing Systems
  • 05:01 Marketing vs Systems Pitch
  • 07:20 Why ERP Projects Stall
  • 11:07 Retyping and Hidden Costs
  • 13:26 One KPI Three Answers
  • 16:42 Forecasting With KPI Trees
  • 20:29 Aligning Sales and Ops
  • 23:47 AI Needs Context
  • 28:34 Office Bottlenecks and The Goal
  • 30:53 Key Person Risk
  • 33:48 30 Day Systems Action Plan
  • 35:26 Where to Find Andy
  • 36:35 Wrap Up and Subscribe

‍

‍

‍

Episode Transcript

‍

Matt Horine: [00:00:00] Welcome back to U.S. Manufacturing Today, the podcast powered by Veryable, where we talk with the leaders, innovators, and change makers shaping the future of American industry.

‍

Welcome to US Manufacturing Today, the podcast powered by Veryable, where we talk with the leaders, innovators, and change makers shaping the future of American industry, along with providing regular updates on the state of manufacturing, the changing policy landscape, and more.

‍

Here's a quick test. Ask a small manufacturer what his win rate is on quotes.

‍

He'll give you a number. Then ask him how he knows, and the room goes quiet. My guest today says that's not a people problem. It's not a software problem. The systems were just never connected. Andy Janaitis is an industrial engineer and the founder of Janaitis Business Systems, where he builds [00:01:00] the systems and numbers function that most small manufacturers can't hire full-time.

‍

He takes the software a shop already owns, connects it into a one set of numbers people can actually trust, and then goes after the biggest bottleneck those numbers expose every quarter. We're going to get into why buying the ERP so often changes nothing, why the same question can get three different answers in the same building, why your AI tools aren't failing so much as starving, and what it looks like to run the goal in the office instead of just on the shop floor.

‍

Andy, welcome to US Manufacturing Today.

‍

Andy Janaitis: Thanks for having me, Matt

‍

Matt Horine: We are very excited to have you on. I love talking about these systems and processes, and especially when we first met, your background caught my attention because you come at this w- from an engineering approach. And in most operations environments, that's a great approach to take.

‍

Oh, starting with the path because the road, you took usually explains how you see the problem. You started as an industrial engineer, then moved into modeling and simulation of computer networks, and a few years as a developer. Walk us through your path. What did each stop teach you about how a business actually runs?

‍

Andy Janaitis: Definitely, yeah. Yeah, so as you mentioned [00:02:00] I'm an industrial engineer by training, and I think that really goes through how I see business and how I've worked throughout my career. Process improvement is definitely, probably even before I became an industrial engineer, that's what pushed me into industrial engineering.

‍

Process improvement is kinda through and through everything we work on. Coming out of school I worked with some government clients for a while around modeling and simulation testing different a- approaches to computer networks. From there, found myself into software development.

‍

Basically, someone on the team had to build a user interface to store the, the model data, and pull up the model data and analyze it. So kinda self-taught JavaScript and and front-end development. And kinda from there, took on a number of different projects.

‍

I've always been working in agencies, so serving businesses both small and large and with a lot of different digital solutions for, all the way from, again, modeling and simulation, data analysis, through to, front-end software even digital marketing in different ways. So I think the thing that the, the thread that goes through all of those is how we look at the [00:03:00] tools we're building and the data analytics how we use those to get a better picture into our businesses, right?

‍

And how to grow businesses based on those numbers. Something I've always been really focused on is process improvement. How do we, get a little bit better each day? I really believe that incremental improvements slowly add up to, one large one large number at the end of the day.

‍

But then especially how do we measure that, and what are the key KPIs? Where should we be looking to tell if the changes we're making are really adding up? A lot of times I'll get that, people will say, how did you come from modeling and simulation into, developing websites into, digital marketing?"

‍

It seems like an odd career path, but at the end of the day, it's always comes bound- back to, how do we improve your business? And underlying each one of those functions has always been how do we measure that change over time?

‍

Matt Horine: Yeah, no, really said. I think it's one of those things that everything is a system. We've talked-- We've had a couple guests on this show. We talk about systems all the time, but the overlap there from marketing to any, kind of operational environment it's true. In 2020, you went out on your own.

‍

What pulled you back towards manufacturing [00:04:00] specifically and towards those manufacturing systems?

‍

Andy Janaitis: Yeah. So I'd worked with a lot of different businesses over the years. Certainly something just from a, a personal enjoyment perspective, kinda getting back to building something, being on the shop floor, kinda seeing some of these cool, unique processes and niches that, that people have developed over the years.

‍

I always love walking into a small manufacturer and say, "Hey, this is, a relatively small business in the scheme of things, but they are, the number one luxury latch manufacturer." So tho- these little types of spots that people have found themselves in. So I love just the people of it and the idea of making something.

‍

But more than anything too it's really, it becomes really visible those process changes that I love to kinda focus on. You can see very clearly as a physical product is moving through a system how does the demand line up to the work we're actually putting against the floor?

‍

What projects are most profitable? Where are new projects coming from? All those types of questions are, some of what I've enjoyed most about digging [00:05:00] into across my whole career.

‍

Matt Horine: Yeah, that's it's something interesting that when we first met you told me something that caught my attention that I think is a really funny thing to put out there for our audience. You told me that pitching manufacturers on paying for leads was like a hard conversation, that marketing crossover into the manufacturing world.

‍

But when you would say your systems don't talk to each other, it got an immediate nod from people. Why do you think that one lands so much better?

‍

Andy Janaitis: Yeah. I think, in general manufacturing, this is, maybe a stereotype maybe some truth, but, it's definitely a little bit older industry. There's a lot of, founders who've been running a family business for a number of years. I think just like in every business but especially in manufacturing, there's some suspicion of marketing.

‍

We've all been promised the world and then, worked with an agency that, that didn't ultimately deliver. So I think from that perspective, marketing just isn't really the pitch in manufacturing. And also, I think specifically in this industry, real deals are made with handshakes and kind of relational marketing.

‍

It's all about your network and who you have relationships with, asking colleagues, even colleagues [00:06:00] at other businesses, how they've handled certain problems. So for all those reasons I think, digital marketing, while there is some gaps there, and honestly, there's a lot of space for small manufacturers to grow their digital footprint I think especially at the top there's just a little suspicion of it and probably they're, they're-- it's not the most pressing need, honestly, in these shops today.

‍

But when we turn to systems, I think that's where a lot of times these businesses, obviously, we all want more business coming in. We want more leads. But I think what we're hearing a lot today is, "Hey, the biggest challenge I'm having is how do I staff the work that I have coming in today?"

‍

It's finding good skilled labor. So a big thing that we're really focusing on is how do we get more efficient with the resources we have? Both, we're not a recruiter. We're not necessarily talking about putting more people on the floor. But with the people that you have, how can we, make their day-to-day go a little bit quicker?

‍

How can we allow your team to take on more things? And we look at that from from the office perspective, but, how can we enable your team to wrap their arms around how your business is going a little bit more cleanly so you're not guessing, you're not [00:07:00] wasting labor on less efficient projects and you're getting the most out of both your people and the software that you're running

‍

Matt Horine: Yeah, really good points to be made. I think that a lot of small, medium-sized manufacturers, the marketing thing is, their collective experiences, it didn't turn out the way that they thought it would, especially with an agency. And, it's a tricky thing to pass, but I think that the systems overlap there makes a lot of sense.

‍

And so let's get into the core of what you do because those systems are so critical. I've run a smaller manufacturing business. A lot of our listeners do the same thing. So I was really excited about this conversation because I think if we have time at the end, we'll talk about my WMS unwind, just for just for the sake of conversation.

‍

But a lot of people listening have lived through an ERP project, and some of them are probably still living through it should catch a f- few folks' attention. You see a pattern with smaller manufacturers, big ERP investment, a big install, and then a year later, the sales team is still running off a spreadsheet and pulling a report.

‍

It takes a week because you have to go ask three different people for it. Why does that keep happening?

‍

Andy Janaitis: Yeah I think a lot of times we have some of these problems, right? So you say, "Hey, we'd like to get a [00:08:00] little bit better picture into how our projects are working or where our sales pipeline is," all the, all these different types of things. And a lot of times in manufacturing specifically, the sell is, "Hey, here's this big ERP that's going to do everything for you.

‍

It's gonna cover sales, it's gonna cover finance, it's gonna cover quoting, it's gonna cover job scheduling." Part of the problem is, yes, it can do all of those features, but ultimately you probably don't need all those features. You might have one of those problems but then you put this big monolithic tool over the top of everything and you find that a lot of, a lot of the stuff that gets set up, because again, you're a small manufacturer, you're trying to minimize the amount that you invest in the setup, there's gonna be a lot...

‍

Either you're trying to do it yourself or you're bringing in an expert to do it, but at the end of the day, a lot of it's getting set up by defaults. So a, a great example, I talked with a custom manufacturer actually yesterday, we were talking about this, and he was saying, "Hey, the biggest issue I'm on JobBoss and the biggest issue I'm having is we get most of our work is one-offs, and every time I go to try to set it up set up the project, it makes me, define the stock and, figure out [00:09:00] how much of this, this product is in stock."

‍

It's like no, this is one widget that I'm making as a one-off for this project or, a batch that's going out. I don't wanna set up a whole inventory tracking system around it 'cause it's just not relevant for this, this particular project, but it won't let me enter the project without doing that.

‍

So these big tools, you get into some of these issues where just by virtue of the size of the tool you... and the fact that it can do everything, you're kinda forced into some of these, these, processes that really are too big for a small manufacturer. So in a lot of cases, the answer isn't, "Let's put a big tool over the top that can do everything."

‍

It's, "Hey, let's find a smaller tool that can do some small parts of it," or, "Hey, you've already been managing this in a spreadsheet, and, honestly, it works pretty well except it takes a little bit too long for somebody to type in the information. Let's try to automate certain parts of that and then connect that spreadsheet to another system where, you can get some of the data in an automatic way so that you can pull the reports a little more quickly."

‍

I think the, the challenge a lot of these, these small manufacturers get into is they feel like they need to go to these very large ERPs and I'm [00:10:00] always surprised walking around some of these conferences and expos as I'm asking about these systems challenges, how often very small manufacturers say, "Yeah, we're already running, JobBoss or Sage or all...

‍

any of these different types of tools." And I kinda take a step back, I'm like, "Wow, man, you're a single founder-driven, you're a sales team of one pretty much you- you've got, pretty small manufacturing floor. Like, how did you decide to put all of this process in place when in reality there's some lighter weight solutions that really would've gotten you there in the first place?"

‍

And then at the end of the day, those big solutions, they're not super maintainable. You've gotta bring in an expert to make any small changes you want in the tool and they're harder to learn. So the people who should be using them, it's not that they're lazy, it's just that it's much quicker for them to do their job in a spreadsheet than it is to learn this big system that they're supposed to be using

‍

Matt Horine: Yeah, that's a really good point, and again, I'm having flashbacks to, you kinda see... i, on my commute to and from our office, I see these, these billboards for large systems or, direct advertising these types of systems. And I know in the back of my mind, okay, it comes with a consultant, with a [00:11:00] ramp-up time to, set this thing up.

‍

Customization, if there is any that's functional, is gonna cost a lot of extra, and so it's a big cost question. I think one other really interesting point you made, you described the integration layer at a lot of the shops as the person actually retyping it into, up to three different systems.

‍

How common is that, and how do you put a number on that?

‍

Andy Janaitis: Yeah. Yeah, we see that a ton. A lot of times, especially founder-driven shops say, "Hey, I'm a salesperson. I'm working out of a CRM, got, managing my prospects, adding information sometimes, maybe not adding some information sometimes, but my team doesn't have access to that system. So once I finally land a project, now I gotta take all that information that I, grew over the nurture process.

‍

We've closed the deal, and now to kick it off, I've gotta get all of that into my ERP or into my project tracking system or whatever system it is that you're using for that." And it's a, a lot of retyping. Some of it can come across incorrectly. There's a cost to, hey, in worst case scenario, we did part of the project and actually have to scrap some real material because, information wasn't added in correctly or we made the wrong [00:12:00] parts.

‍

But even in the best case, there's a time cost associated, even if it works perfectly, just the, the cost of, having to copy that stuff over, and the wait time where, your lead times just necessarily grow because, of course, you close the deal on the busiest day, and then it has to wait for a week just to get some of that founder or salesperson's time to go ahead and do the kickoff that really should be an automated process that doesn't need to go through.

‍

So that's just one example, but we can see that across, a number of different systems. I've heard similarly like, "Hey, invoices are all manual. We'd run the whole project. We've got it really dialed in. The project goes through the shop floor, but then at the end of the day when it's time to cut the invoice, we've gotta, get that information out and type it in, get it set up."

‍

And again, same problems. There's a time cost with doing that integration and getting the in- invoices that really should be automated. But even beyond that, there's this risk cost that, you know, 2% of the time you're gonna, you're gonna fat finger something and the invoice goes out and then you look less professional to the client and all those different types of things.

‍

You're either overcharging or undercharging. Just [00:13:00] thinking that in mind that there's a, the real cost as, as well as, this additional risk that gets added on just by, humans make errors in these types of processes.

‍

Matt Horine: Yeah, that's a, a really good point, and I think about the amount of times, and it's not even like a change order. Like there's not, there's nothing dynamic about it. It's like a misquote, so you're left, for lack of a better phrase, you're kinda like pants around the ankles.

‍

You've overquoted or underquoted and have to go back to the customer and say, "Actually, it's more." Or, you're upsetting the customer because they don't think they're getting what they're paying for.

‍

So yeah, it's a really good point. I wanna go back to that win rate question from the top of the show because I think everybody listening has been in that meeting.

‍

You're you had, we talked in our when we met each other and you told me kind of the story. You ask for the quote win rate, and you get one number from the CRM, get another from the ERP, and a third from a spreadsheet. Why does the company, or why does the same question get three different answers, on, on the quote win rate inside of a company?

‍

Andy Janaitis: Yeah, and I think that's a great example, and we see that, it's just one example. We see that across a lot of things. But it's, a lot of times that's where we start, right? Because we wanna measure, how are projects coming in, how [00:14:00] healthy is the demand for your services, and then, ultimately how profitable are the projects that come in.

‍

So we always roll up a few key, key KPIs, and that's what we're reporting on, that's what we're helping the business to kinda track and grow to. And of course, win rate comes up big in there, right? It's, okay, here's the pipeline. Here's how many, people filled out a form on your website, or how many people the founder met at a conference and added their information in into, to HubSpot or whatever.

‍

And we get to this thing where we start to think, okay, quote win rate. Just taking this example, yeah, I've had this conversation with a founder where we'll say if I pull it from HubSpot it looks like I closed about 10% of my deals." But really, I think the rate should be a lot higher because, half of those deals weren't real deals to begin with.

‍

They were spam forms that came in. Or, hey, there's actually, kinda moving in the other direction, there's some wins on there that never really got tracked as open deals in the first place 'cause it's just, revenue that maybe never even made it into your CRM because the, the founder, had a conversation within a day or two, made the sale, made the quote and just entered it directly into the project [00:15:00] tracking system.

‍

It never really even hit the sales system. So that's like from a founder-driven perspective, you can imagine, with a key salesperson, all the same types of problems would happen. Even one person in one system, there's different ways to think about it, and as you pull the numbers out, their immediate go-to is, "Okay here's the number I pulled out, but now how do I adjust that with some of my, tribal knowledge of how different things work?"

‍

Another great example though is when you actually have two different systems, so we see that a lot, where, hey, you can have, the revenue that you pulled or the amount of deals that you actually closed. That's gonna be very different if you look at it in your CRM and look at how many closed deals there are that went through your CRM pipeline.

‍

I could almost guarantee if you look at that versus going to QuickBooks and looking at your financial software and say, "Okay how many first-time invoices got sent out?" It's gonna be a little bit different. And, even at that, that's probably an example of something not getting recorded or something getting kicked off.

‍

But oftentimes, you even have some of those questions about accrual versus, you know, cash. Like, when did the invoice actually get paid versus when did the project get completed? So when do we call the actual [00:16:00] win? Those are some of the things where the data's there, but there's maybe a little bit different definition of it in different systems.

‍

Whether you call the quote one the day that you send the first email that says, "Hey, let's do this," versus the day that the invoice goes out versus, the day that the project work actually begins. Whatever it is, just clarifying some of those numbers so that as your sales team gives their report, that's gonna match, the numbers that your CFO is looking at, and that's gonna match, the numbers that the production team is looking at as well.

‍

Matt Horine: Yeah, that's a, and I think I caught a, a keyword there about the invoicing and the actual dollar cost. And, this gets into where it's the money you can't see. The part I think that probably hits closest to home for operators who are listening where the money actually leaks.

‍

And you had told me if you don't have a forecast, you really don't know whether you're having a good month or a bad one. How should a small manufacturer set those monthly targets? This is a little bit more systems-focused now, but how should they set those monthly targets, and how do they use them as the month goes on?

‍

Are they dynamic? What what about this is missing from a lot of small manufacturers?[00:17:00]

‍

Andy Janaitis: Yeah So I think we all have a habit of doing this in a, I'm running a small shop myself, so I my team, it's, I'm doing a lot of the sales myself. So a lot of the times as I'm thinking about forecasting, thinking about, "Hey, what's revenue gonna be at next year?" I'm running a constant, running total in my head of, okay what was revenue last month?

‍

Plus what's recurring? What new projects are coming in? What, what has a chance of closing? A lot of times founder-driven shops are doing a lot of these calculations in their head, but it just never really hits the system. So I always say kind of the first thing is, roll up and just define what those numbers should be.

‍

So again, if we're gonna call it booked revenue, just like we were just talking about make sure you have a clear definition of what that means. Is that invoiced revenue? Is that deals that are completely, we have complete sign-off on, or is it, hey, contract signed or what- whatever it is, just getting to a point where you have a really clear definition of what you mean by that metric of, say revenue, say project capacity whatever the metric is that we're trying to track to.

‍

But then from there, when we get to forecasting, this is where we wanna start to look at ultimately it's [00:18:00] forecasting. We're trying to take a guess at what that's gonna look like in a given month. So I always like to start from a perspective depending on the shop, and it depends on whether you're whether a better anchor is last month if you're growing over time or whether a better anchor is this time last year, so you're looking at year-over-year numbers if you have a little bit more annual, a business cycle that you've got to take into account.

‍

But what I like to do is say, "Hey, let's start from the baseline." Like where were we last month or where were we year over year last year? How much do we want to be growing from that? And the number, the percentage that you wanna be growing is typically gonna be, based on what you've been doing historically.

‍

So if you're looking at what's our month over month growth over the last couple of months? Or what's our year over year growth over the last three months? How did they grow over last year's prior year? Apply that to your baseline metric and start to get a feel for okay, that's where we're, where we want to be this month.

‍

We've been growing, about 30% year over year. So this month I'm gonna look at last year's figures for this particular month, apply [00:19:00] 30% to that, and that's where we want to be this year if we want to grow 30% over the course of the full year. And then from there you just start breaking it down.

‍

I always like to think of this as like a KPI tree, right? So if we talk about revenue as that top line going back to the goal, one of our, one of my favorite books, it's like the purpose of every business is to make money, right? Revenue is like that's what we're trying to do here.

‍

Realistically, it's actually profit, not even revenue, but, if we're measuring revenue as a way to measure how the business is doing in a given month. From there, you've got one revenue number that probably breaks down to how many projects and average cost per project.

‍

So now you've got two extra KPIs you can break down. Then number of projects probably goes to how many propo- Proposals went out and the proposal, the quote win rate, and you've got two extra KPIs. So you can kinda get from the big KPI that matters, the big forecasted number that matters, what are all the intermediate KPIs that we have to hit this month to make that top line number actually work?

‍

And now you can start to track and say, "Okay how many new quotes have to go out by day 10 of the month, by day 20 of the month, if I'm trying to hit a total of 50 new quotes over the course of the month?" [00:20:00] Those types of questions. The big idea there is just to be able to say, hey, every or most manufacturers we work with can tell you at the end of the month or a few week- days or weeks after the end of the month, how did last month go?

‍

Everybody closes their books at some point and figures out, what was the revenue and the profit. But it's nice to be able to see on day 10 of the month are we tracking towards where we really wanna be or not?

‍

Matt Horine: Yeah, that's a really good point because a lot of smaller manufacturers, it does take them three weeks up to three weeks to close out their books, and there's some costs, and decisions that can be made too late. One, one other really interest part about this is the dynamic within manufacturing organizations.

‍

We had a guest a couple weeks ago, Josh Long who wrote "The Bottleneck Breakthrough." I actually think you guys have, a lot of the same viewpoints on this, and our conversation focused a lot on the dynamic between the commercial side of a manufacturing business and the operations side. How do the connected numbers that you're talking about, when those systems actually connect, how does that change the conversation between the ops side and the commercial side of the business?

‍

Andy Janaitis: Yeah. It's a lot of understanding, right? It's, again I, kinda reference the goal. [00:21:00] Everybody necessarily, especially as you grow, you have to Everybody has to specialize in their little fiefdom. Like how many how many projects are we moving through our station in any given day or week?

‍

And that's part of that KPI breakdown is figuring out what are all the numbers that need to line up. But if you start to find places where, hey, we can, for instance, if we're talking about looking at revenue or, hey, sales team says, "Hey, we can get more projects coming through," if we're light on capacity, we can start getting more project coming through, but they're not particularly profitable projects, that doesn't really help anybody, right?

‍

So when you start to get to every individual department or every individual station really optimizing for the thing, the one number that matters most to them, you can get yourself into some bad situations because those optimizations aren't always what rolls up to optimizing for profit for the entire shop.

‍

And I think a lot of what we like to do in connecting systems is help to make those pictures a little bit clearer, right? So if it is connecting different types of projects, again, think a sales team [00:22:00] trying to, get as many projects through as possible. If they're stacking up a bunch of projects that are gonna require certain processes that are over capacity right now, yeah, they've sold a bunch more work, but we don't really have the capability to produce it and to do it.

‍

Even if it's a little bit harder for them to sell revenue or sell projects in an underutilized project, maybe that's a better case for us. So can we roll up the system so that sales team can see immediately where there's capacity and what we should be selling more today, as opposed to having to wait until the end of month meeting where they start to hear, "Hey, next month, let's start to focus on selling process A instead of process B."

‍

Let's try to see that in real time and say, "Hey, next week, process B is gonna have some open capacity. Let's really work on closing some of those deals that are in the pipeline there." Even if they're harder to close, it's a better use of our time than closing a bunch of process A that we don't have capacity to start on for another two months anyway.

‍

Matt Horine: No, those are really great points. And again I'm back to the flashback section. When I was a, this is several years ago now, I was a an operations manager [00:23:00] in a business that did it was a SaaS model, but it was also this, this hodgepodge like hardware as a service concept.

‍

And the sales commercial side would go sell these projects based on the software model, their seats and their equipment. The commitment to hardware was significant, and it wasn't those streamlined processes where we'd have this hardware integration, and there was like this payback model on the lease of the hardware.

‍

And it was brutal 'cause you would overcommit on the procurement side. You would buy way too many parts and pieces where you basically had this massive payable to something that hadn't even been implemented on a project yet because of lead times, because of whatever, the constraint was.

‍

So as you talked through that, I was just remembering, back maybe a decade ago, how the sales guys overcommitted and, the hardware piece of it was very difficult because it wasn't an allocated set, can't do just in time when you're building custom solutions. So I... that's a really good real world experience e-explanation there.

‍

But let's shift into the artificial intelligence segment of this and talk through it. It's a, it's... I wanna talk about it the way you do without the buzzwords. We had a really good, intro conversation about this and you told me about companies that bought, stacks of [00:24:00] Claude seats.

‍

They tried writing emails, it didn't like what came back, decided AI just isn't there yet. Your answer was that it's not really the Claude problem. It's not really the AI problem. What was

‍

Andy Janaitis: Yeah. So it's a great topic of conversation, and honestly, that's what everybody wants to talk about right now, right? Is it's like everybody's feeling that crunch of, are my competitors gonna leave me behind if I'm not implementing AI? I can't tell you how often I hear I've gotta do something with AI.

‍

I don't know what I'm trying to do. I just, I gotta do something with AI because I know it's really important right now and we gotta figure this

‍

out." And I, just as often I hear people at that exact example is, "Hey, I paid for Claude. I tried to get it to write a couple quick emails, and honestly, it's just not there."

‍

Probably in a year we'll pick it back up, I think that the tools have a little ways to go. That's a refrain I hear all the time, and the answer isn't... and I can tell you because in our own business, we're using AI really top to bottom across all of our processes, and for the right projects, it is there.

‍

We're getting a lot of really high-end work through increasing our capacity to a point that we couldn't have done previously, and the answer is connecting the AI to the context of your [00:25:00] business. So the way I like to think of that, so first and foremost, if you wanted to write emails, if you wanted to write quotes, connect it to give it, feed it the quotes that have already gone out.

‍

So you're starting from a point of, "This is how we like to do quotes. This is how I like to write emails," both from a voice perspective, but also there's differences in your business. Even if you have an AI-specific quoting tool or an AI-specific email tool, it's generic in different places, and there's specifics to your business that aren't gonna be captured there.

‍

So the number one thing we gotta do is connect the underlying systems so that before the AI gets started on whatever project you're gonna give it to, it has some examples of how you did that project in the past. And those connections usually aren't that hard. Most of these tools that we're using these days, whether it be your quoting software, whether it be your ERP, your CRM your Google or Outlook for email, all of them have easy connectors to get all that data over into Claude.

‍

But then the thing that I think is next most important and what often gets overlooked is, yeah, the first time you test it out, and this is why a lot of [00:26:00] AI pilots fail, is you run through some tests and you say, "Eh, it's 90% of the way there, but it's not 100% of the way there. I couldn't send this to a client, so it's not gonna work."

‍

What you've gotta do at that stage is say, "Okay we're gonna take it, take that draft that's 90% of the way there. I'm gonna bring it into my email system." Again, it's all connected, so I can do that very quickly and easily. It's not a lot of extra work for me. I'm gonna make all the edits, all the drafts that I, or all the tweaks that I need to make so it's perfectly in my voice.

‍

It perfectly answers the customer question. It's ready to go out, and then I'm gonna send my version, right? So you say, "Oh that was a lot of extra work. I-- You still had to do a lot of editing there." But the crucial step here is now we're gonna let Claude go ahead and diff those two versions together and say, "Hey, this was my draft.

‍

This is what you ultimately ended up sending. Let me figure out what the differences were there and learn from it," and start creating a set of rules. And every time you do that, it gets a little bit smarter. It gets a little bit closer to your voice. And before long and honestly, not very long at all, we start to see those types of self-learning systems will get to the point where It is almost customer, right?

‍

I do believe a [00:27:00] human should be in the loop almost 100% of the time. I don't think you should be sending something out direct to customer that was written by AI without even looking at it. But, I find oftentimes, I'll have AI output that I review and maybe I have one or two tweaks, but it's, pretty close there based on all the work we've done, over, over the last year or so in training the models to really hit our very specific function that I'm asking it to do in exactly the way I'd like it to be done with my voice

‍

Matt Horine: Yeah, that's so critical. I think that people I think the only word I can come up with is something like abdicate. They think that it's just going to do this for them, and we've had so many conversations about this on our show, and I... you do this every day. I think you summed it up if I catch it right, AI comes last, in this process.

‍

It's not necessarily like a pilot to in- to go explore new systems, if it'll do something for you. It's more of a pull it all together and enhance, process and efficiency if that's, if I'm getting all that right.

‍

Andy Janaitis: Yeah, I think you, you've got it there. It's like you're, you're-- you, especially the first couple times through any process, you don't wanna be pointing AI at something that you've never done yourself kinda [00:28:00] thing. You wanna have, somebody who's really an expert in how this should be done, whether it's quoting, whether it's the emails that you send, you wanna be in your voice a- as you're the expert there.

‍

Whether it's, pitch decks, that's a, a big thing. It's like, why would you think that AI's gonna be able to create a pitch deck and then go out without somebody who's actually an expert to be able to review the system and say, "Hey, this is how we, as a company over the last 10 years, have decided we do pitch decks."

‍

I think those types of things it's yeah you wanna make sure you really understand the process yourself first and then point AI at it and get a lot more efficient in how you do the process, but never taking the human 100% out of the loop.

‍

Matt Horine: Yeah, said. Let's go back to the goal because I think that's something that most operations professionals have probably read or at least had somebody quote it to them on a shop floor, right? You're taking that idea somewhere a lot of people don't. At any moment, one constraint limits the whole system.

‍

How do you apply that to a, an office environment or to the office, and how do you decide which bottleneck to go after first?

‍

Andy Janaitis: Yeah. It's really important. I think that's, So first of all, yeah, "The Goal," if anybody [00:29:00] listening hasn't read it, go ahead and get a copy of "The Goal." As an industrial engineer, near and dear to my heart. That's probably my North Star of how I think about a lot of these problems.

‍

But at the end of the day, this whole concept, and sometimes people will push back and say, "Okay you're telling me about building a dashboard where I'm seeing my key KPIs, and then you gotta build a forecast for each thing. Feels like a lot of work here. Like, why don't we just do the work and then, if we're doing good work it'll all end up well in the pa- in the end."

‍

And I think the, one of the biggest values we get out of this whole this whole activity of creating, measuring how we're doing today, and then creating the forecast for how we wanna do in each mo- in the next month, is we can start to see where those numbers start to break down, right? So if we say, "Hey, we missed revenue or we missed our revenue target by 20% in this, this given month," let's look at that and break down, was it that we didn't sign enough projects, or we did sign plenty of projects, but our project size really decreased by X percentage, or hey, we're able to sign the projects, but we weren't able to get them out the door because we didn't have capacity in something. If you've broken down, here's your top level goal [00:30:00] and here's all the different KPIs that you need to hit to get to that top level goal, as you're going through the forecast, it becomes very clear that typically you're not gonna miss on all of the KPIs.

‍

You typically are gonna miss on one or two, and that's where you start to say, "Okay that's my foc- focus point. It's, we're closing on volume, but we don't have capacity to get the work done. So okay, let's look at the shop floor and figure out what we can do there. Maybe it's that we can close similar volume, but on better types of projects to match the capacity we have.

‍

Maybe it's, we can do, do fewer projects that are a little bit larger size. Yeah, overall our revenue might be a little bit less, it's gonna be easier to manage on the floor, and we're gonna be able to get things through a little bit more quickly." So it's a process. But those answers are gonna be different for every business, but the, the key piece is, finding a methodology so that you can look at not only how you're doing at the top line, but what are the numbers that roll up to that top line, and where can we find, room for efficiencies and where we're falling down at any given point.

‍

Matt Horine: Yeah, that's something comes to mind, as we're talking through that. One of the biggest constraints in a lot of operations is [00:31:00] that one person is that system. We all know it. We all know who that person is. We've seen it before. It's because of, it works, everything works 'cause of who they are, and it stops, the week they're out.

‍

How do you fix that without making that person feel like they're being replaced? That's something that people are very sensitive to. We've talked on a lot of touchpoints on what validates that, the, our intelligence conversation driving some type of efficiency, and somebody holding the keys that makes them feel valuable.

‍

But how do you do that without making them feel like they are being replaced or they are being marginalized?

‍

Andy Janaitis: Yeah. So a couple different answers or ways there. I think a lot of times it's a, it's an easy sell because that one person, especially in small shops that one person comes down to the founder, and it's just making the case of, "Hey, how do we get some of this off your plate so that you have a little bit of extra time, a little bit of breathing room to really run the business as opposed to working in the business every day?"

‍

But as we go down, that key person risk still lives, in different departments, and we do see that all the time that some of these processes live with a person. And it can be a challenge, especially somebody who's built their whole career on, running, knowing this system, understanding how to solve this problem.

‍

You start to say, "Hey let's take that [00:32:00] off your plate,"

‍

Then it's like what am I-- what's my value here?" I think the big thing is that's where keeping a human in the loop is really important. One, I think as long as you're communicating to them that, "Hey, the knowledge that you've gained over all these years is so important that you are the leader here who's gonna help guide how this process should run and help us to understand as we make it a little bit more efficient."

‍

We don't wanna just, again, we don't wanna just set AI onto a particular process, say, "Hey, AI's gonna run all our invoices now," and then just set it and forget it and let it go. You're gonna say, "Hey, we wanna take some of the stress off your plate. We wanna take some of the time off your plate to help you to do higher level thinking.

‍

But at the end of the day, you're the one who knows all these little rule sets and all these different things. You're gonna be the one who's really in charge of helping the AI to get to a point where it can work serviceably to do some of those things for you under your kinda discretion and control."

‍

So I think it's just framing that as like we're never replacing somebody, we're helping them to do their job more efficiently so that they can apply their knowledge to [00:33:00] higher level problems. Typically we don't see that key person risk showing up with somebody who's not already overly capable.

‍

The reason they are that key person that everything flows through is because they're the people who can knock everything out and can do everything. Letting them know that "Hey, you've become a constraint because you're so capable. Let's take some of the menial work off your plate so that you can, really elevate the impact that you're making today."

‍

Matt Horine: Yeah, no, it's y- it, I've heard it phrased as extreme competence at, as a very specific task or system, and that's what it's all about, is freeing up that capacity. If that person is that talented to be doing that and holding it all together, imagine what they could do if they were, unshackled from the burden of dealing with those types of things.

‍

And people put that in place to, describe their own value sometimes. That's not necessarily the case. Your value was you built it, so now go do whatever that system was designing, your team to go do or enhance your capacity. Really great points. The, this is the part of the show where we take a billable hour from you or something to that effect, and we call it the free advice, you know, section and something that's actionable for our audience.

‍

If a manufacturer wants to start looking at it through this [00:34:00] systems-oriented lens or thinking about the types of things we talked about, in the next 30 days, what's the one thing they should do?

‍

Andy Janaitis: Yeah. Number one thing is define the key KPIs that, make your business run. So understanding, obviously everybody knows revenue and profit, but start to break that down and understand what are the underlying numbers that, that really roll up to you had a good month or you didn't have a good month.

‍

And then start walking through a, a what if. If I wanted to track all of these numbers in real time every day, what would that cost me? What, how would I even go about doing that? What people would I have to pull into the room to pull different reports? And from there it'll start to become pretty, pretty clear pretty quickly.

‍

All of this is very feasible except for there's one system that no one can get the data out of the ERP or, no one can get the data out of the financial software. So that's where we start. Okay let's try to figure out how we can make it a little bit easier to pull that, that data out.

‍

What, how can we connect that to another reporting software that'll allow us to pull the numbers together from two different systems? But just start by writing down like what's a good month look like to you? And start to figure out why you don't know how you're [00:35:00] progressing towards that number today, or, like immediately in real time right now.

‍

Those are your bottlenecks to start solving and start figuring out, how to get this, this data a little bit more clearly for you.

‍

Matt Horine: Yeah. Yeah, that's I think so important. A lot of times people just look at the QuickBooks report, especially like a small manufacturer. They're just, "Hey, did we have a good month or bad month?" Or they're hoping for the cyclical cycle part of their business where it's like, "Oh fourth quarter's coming up, so everything's gonna be great."

‍

What if it could all be great? Start asking those tough question what ifs. If somebody wanted to, get in touch with you or talk about some of those what ifs, where can people find you? What's the best way for someone to take you up on that first conversation?

‍

Andy Janaitis: Yeah, so you can always find me on LinkedIn. I try to be relatively active there or come to the website janaitisystems.com. It's my name systems.com. You can book a 25-minute call with me. I'm more than happy to to hop on, get to know your business a little bit and help identify what is that first key win and where can you really save some time, money and get a little extra clarity.

‍

We u-usually get a lot of value out of those calls. We're also gonna be running a webinar on November 19th. I don't know if this will [00:36:00] hit before then, but if you come to junaitisystems.com/webinar, you can either sign up for that one or if it's gone by, we're, we have a webinar series where we're constantly putting out new ones so you can sign up for the next one.

‍

But but a webinar kind of going through how we think about these problems, how to think about the individual constraints and get from, hey, the systems just don't talk to each other, to what are some of the underlying issues and how do you start to resolve them in your specific

‍

Matt Horine: Excellent. Andy, this was a great one. We'll be sure to link all of those in the show notes. I think we're gonna-- this will hit the air before that, that November webinar we'll make sure to link all of that. But this has been a great one. Thank you so much for joining us today.

‍

Andy Janaitis: Thanks for having me

‍

Matt Horine: Again, Andy offers a free 25-minute call for any manufacturer dealing with these, these kinds of headaches.

‍

No pitch, just a walkthrough on what's going on with your systems, and you can book that at genitussystems.com. We'll link again in the show notes. Thank you for joining us. So to ahead of the curve and to help plan your strategy, please check out our website at veryableops.com and under the resources section titled Trump 2.0, where you can see the framework around upcoming policies from the administration and how they will impact you as a manufacturer. If you're on socials, give us a follow on LinkedIn, X, or Instagram. And if you are enjoying the podcast, please feel free to follow the show on Apple Podcasts, Spotify, or YouTube and leave us a rating, and don't forget to subscribe. Thank you again for joining us [00:42:00] and learning more about how you can make your way.