Episode 66 Transcript

Insights into AI Transformation in Professional Services w/ Ken Ringdahl

    Banoo Behboodi: Hi everyone. Thanks for joining our podcast, Professional Services Pursuit, a podcast featuring expert advice and insights on the professional services industry. I've got to say that one of the things that I love about working at Kantata is the amazing people that I get to work with every day and the awesome executive leadership team. And I'm joined today by one of those executive leaders.

    We've asked our CTO to join us on the podcast, Ken Ringdahl. Ken is a technologist with over 25 years of engineering experience, specializing in enterprise infrastructure software. So the topic today is going to be exploring Ken's insights on AI, its impact on professional services, and some of the does and don'ts and considerations as you're looking into implementing AI within your professional services business.

    Thank you so much, Ken, for making time and joining us. I know you're very busy.

    Ken Ringdahl: I'm really happy to be here and excited to talk about one of my favorite topics.

    Banoo Behboodi: Awesome. Let's start with you telling the listeners a little bit about yourself and what you do at Kantata and life before Kantata, please.

    Ken Ringdahl: Sure. I'm a software engineer by trade, and I got a computer science degree in college, became a software engineer, and worked for several small startup software companies. That really gave me an opportunity early in my career to wear many hats, dig into many things, and work in a very agile and quick environment.

    Then that brought me into larger executive engineering management roles, similar to the role that I'm in now. But I consider myself a technologist—a builder per se. I love building software technology as one of my passions in life, and now that I'm the CTO of Kantata, I'm responsible for our day-to-day engineering, our service delivery, global support organization, and long-term technical strategy.

    Banoo Behboodi: That's amazing. Thank you for sharing that. Now, I know there's a lot of podcasts and discussions around AI, and I know we've done a few sessions. I know there is a lot of terminology as well that is being thrown around, around AI, generative AI, what the impact is, and what all of this means in the tech space.

    What are the key differences, specifically in the PS technology realm, that the professional services team needs to be aware of?

    Ken Ringdahl: Maybe I'll start by giving us all a very quick foundation here. So because there is a lot of noise around AI, there's a lot of confusion around AI. It's been around for a really, really long time. Certainly my entire career. AI has been around in different fashions.

    There are all sorts of different types. There's things like computer vision, and this is like facial recognition and identifying logos when someone tries to impersonate the logo of a major brand that you may get, or phishing email that's got an Amazon logo that's trying to get you to click on an email and steal credentials. That's computer vision. Robots and things like that in the Amazon warehouse. Machine learning—we'll talk more about that today. It's something that's very sort of front and center, especially for PS firms, around data and making decisions on the data. Natural language processing—I think everyone engages with some form of NLP regularly because this is about virtual chat bots and spam detection.

    But when we talk about generative AI, Gen AI is really almost like taking the air out of the proverbial AI realm in the last 18 months. It was probably late 2022 that Gen AI really sort of took off, and the technology hit an inflection point. Gen AI is named as such because it generates content.

    So if you think of something like writing a letter, providing a summary of a book, or drawing a picture, there's some output, and that's what the whole notion of generative AI is. We're really in the very, very early stages of generative AI. For those who know the Gartner hype cycle, basically all technology and innovation run through this sort of cycle.

    People can't see what I'm doing with my hand, but it's almost like a sine wave. The technology in AI is very similar to generative AI. Initially, a technology comes out and people think, “Oh, that can solve so many problems.” So, the hope of what a technology can do hits a peak, and then you lose confidence in it because it can't do everything it thinks it can. Then it comes right back up again.

    Actually, I checked the Gartner hype cycle for AI, and the last time it was published was in August. Literally, generative AI was at the very peak of the very beginnings of what they call inflated expectations. So we're really going through this process, and it will go through just like other technology will go through here.

    But I would say that generative AI has hit that state almost faster than anything else that we've really ever seen. So there's just a ton of noise and a lot of aspirations for generative AI, specifically for PS organizations.

    Now, what I'll say about PS organizations is that, PS as we all know, it's a lower-margin business. It's a people-driven business, and PS relies on the productivity of people. As a result, there's a lot of benefit from the productivity gains that can come from generative AI. So there's just the generative AI and the data analysis; there's just so many things that PS can benefit from. And I'm sure we're going to get into more of those as we go through here.

    Banoo Behboodi: Like you said, it's people-powered businesses, and ultimately, it's not just that they need to be efficient. They need to be efficient because they bring a skill set and knowledge to their customers. That is much more valuable than some of the administrative work that gets consumed.

    With that in mind, some specific applications based on your sense of the benefits that AI can have for professional services will be discussed more immediately as we work through the hype.

    Ken Ringdahl: Yeah, absolutely. So as I said, we're kind of in those early days of generative AI, but what I would say is that what Gen AI has proven to be really good at in the early days is executing tasks and increasing productivity, especially for what's known as knowledge workers. So there's things that really allow knowledge workers to focus on things that are really high value and take things off the table that are lower value, or more like remedial tasks.

    So think of it as a little assistant that can handle things. So I love data. In preparation for this, I've pulled together a little bit of data to just give people an idea. TSIA did a quick poll, about six months or so ago, of PS organizations as sort of primary focus areas for AI development and planning.

    At the very top of that list, two-thirds were data analysis and insights, and right behind that was automating or digitizing operational processes. So, really, looking at your data, everything else was below 30%. So these are the two areas that PS firms are really digging into, and then everything else is sort of secondary. But it also highlights how important data is and the value of data engineering to PS organizations.

    I'll take the flip side of that, and we'll talk about a few opportunities here. On the flip side, improving service delivery was a really low priority for PS firms. It brings forth that the core of PS organizations as delivery is operational excellence. So it's not surprising to see that initially PS firms are focused on, “Hey, how do I improve my own operational processes, and then looking at how do I leverage AI in a more customer-facing fashion on the service delivery side? So I think there's some really quick wins in leveraging AI for operational efficiency.

    Some of those opportunities, and I'll bring a little bit more data here. I was actually really surprised. This is actually some data that we also, as Kantata, surveyed, and many of our customers and prospective customers too have an average of 31 days to deliver a proposal. I was a little bit floored by it. It's taken a month to deliver a proposal.

    Again, coming back to the notion of Gen AI, some of the things that it is really good at are content generation. So think of generating a proposal, and probably that proposal will be even more optimized because the next piece of data shows a 54% average success rate on proposals, i.e., the proposal was accepted, and you have a deal when you have a client going forward. So AI can really analyze historical data and optimize your proposals to drive a greater success rate.

    Then lastly, very related to this, after you win a proposal and have an engagement, you have 18 days to staff a project. There are so many opportunities to optimize resource forecasting and automate project staffing that I think there's a large opportunity for AI to really influence that and really take all three of these metrics and drive them up substantially and drive better outcomes for PS organizations.

    There is a lot of enhanced efficiency, increased creativity, delivering superior service, and really increasing those outcomes for customers, because the other side of this is obviously repeat business. A big driving factor for PS organizations is if you deliver a good project and a good outcome for that customer, and AI can absolutely help improve that and drive success for your business in almost an exponential fashion.

    Banoo Behboodi: I love that. So it's not just about the revenue and driving the revenue and margin, but also enhancing the experience and the repeat business that comes from customers, and just curious to extend that, the impact that you think it's going to have on the employee and the talent experience from your sense.

    Ken Ringdahl: There's no question. Maybe I'll even step back and talk about our product vision from a Kantata perspective, because I think it plays into it. Of course, I assume everyone that's listening to this podcast understands Kantata is a vertical SaaS player. We focus on service organizations. We've been largely focused on what I was just mentioning as the operational excellence part of the operational nature of PS organizations.

    Over the last year, we've looked at our customers and our target audience to see how we can help them in their businesses. And so we started to branch out from that. You mentioned employee and client; those are two areas where we have some features and functionality in our products that service them. But it's much more important for us to really look at that as almost a three-layer cake.

    The operational experience, the employee experience, as we know, retaining employees and growing those employees and training them, and then the client experience—we touched on a little bit of that just a moment ago around happier customers leading to repeat business.

    There's lots of opportunities there. And specifically, like resource management, skills, and other areas, AI can provide so many benefits to help drive those other areas outside of the operational part of the employee and the client experience.

    Banoo Behboodi: I love it. So, Ken, obviously, there's a lot of hype around generative AI, and everyone is trying to figure out how they can utilize it in the best way to address their customers, business, and talent. But that doesn't come without challenges and risks, as businesses and professional services targeted here for our discussion explore how they want to apply generative AI.

    Can you walk us through some of those challenges, some of those risks, and the best ways to mitigate and address them?

    Ken Ringdahl: Yeah, absolutely. So I think there's probably two ways to look at this. One is, “Hey, how does the organization get in the game?” And then, when they do get into the game, what are some of the risks associated with that? So maybe about getting into the game, and we hear this, our customer base, we go from small agencies all the way up to really, really large full-service, professional service organizations. Those larger organizations have the budget. They have needs. They have the mechanism to dip their toe in the water to try some things out.

    But I would say that on the whole, like a lot of PS organizations, they're not really technology companies. They are a consulting organization. They really need a lot of help and guidance along the way to get started.

    I would also say that there's this notion of potentially being left behind. When innovation comes forward, those who adopt and embrace this innovation are going to have a fundamental advantage overall. So, I think helping PS organizations really gets started.

    Coming back to the data, I mentioned that I love data, and our survey showed that so far, 41% of PS organizations have yet to identify a budget source, which really means, “Hey, could they try a few things out?” Sure. But they don't really know the way in which, if they decide, they want to move forward with how they would do that. So I think there are some inherent challenges associated with that, but they also need guidance. And I think there are some things that we can provide to PS organizations to help them maybe narrow down their focus, which would then help them identify the budget. So, that's as far as getting involved.

    I think the other part of this is some of the risks, because I think we've all sort of seen that some of these things have gone viral around some sort of generative AI chat bot kicking back incorrect answers or, even worse, some other things.

    But I'd say there are a few things. One, the accuracy of the data is really critical. What I'll say about generative AI is that a lot of it is this whole idea of how you get out of what you put into it. This whole notion of what's the best way to input into a chat bot or into a large language model. To a certain degree, the accuracy of your data is going to reflect the accuracy of what you provide, and so I think that's certainly an important concept there. And, as I said, some of these failures have really gone viral. But that's improving over time, and these things are only going to get better.

    We are in the very early stages of these, as we mentioned earlier, probably 18 months really, since generative AI has really sort of hit this inflection point. I think this is an area where technology is going to increase at a very rapid pace over the next two, three, or five years. So, the accuracy of the data is an important one.

    Privacy and security are another one that's really, really important. A couple of things. You should assume everything you put into a public model is public. When you're using a public model, you're helping whatever that model is to train itself. So everything that you're inputting there goes in. So that should be something everyone knows. That's there.

    There's a compliance and legal risk associated with that because, with the data that you're putting in, you need to understand whether that is confidential or protected information. So there's a compliance and legal risk thing there. What PS organizations can do, or what any organization can do, is really look to leverage enterprise models.

    This is where you very clearly get what you pay for. You really need to invest in a good enterprise model. It's going to cost more money, but you're going to get better security and better privacy as a result. Quite honestly, for us at Kantata and for other enterprise software companies, it's really non-negotiable. That's the path you take. Our customers will demand it. We are going to demand it from our own perspective.

    Then, coming back to a point I touched on earlier, there are inherent risks with actually doing nothing. I mentioned this sort of competitive balance. If you're competing against someone else that is leveraging AI and new innovation, it's going to be very hard to compete now and especially in the long term.

    So I think helping our customers and helping PS organizations get started is really, really important to us. And I'll just say that, like those who have over time embraced innovation, embracing technology almost always results in a good outcome. You don't have to be on the really, really early adopter side, but I would just say history has shown those who bucked the trend or pushed back against technology innovation end up losing in the end.

    So what I would encourage PS organizations to do is find out where you can make an investment, find out where you can learn and understand more, and set up a plan to get involved.

    Banoo Behboodi: One of the key things you mentioned is the fact that quality data is golden, and to get to quality data, I think all the aspects of people, processes, and technology have to be in place. So we're rethinking this whole spreadsheet strategy, where we're capturing data, and it's working like spreadsheets. People love their spreadsheets. So they're getting adopted.

    But ultimately, it's not strategic because it's not getting you to where you need to be to be able to push the dial in terms of efficiencies, operational, etc., when it comes to generating AI.

    I love the fact that you stress quality data, and it's impossible to do that without proper processes and technology to support the collection of that data.

    That being said, how do you then implement it? With that point in mind, what are some of the considerations for customers? I am assuming that depending on where you are in your maturity, the path you take is going to be very different. Again, if you are spreadsheet-oriented, it may be that you have other considerations before you get to generative AI.

    Ken Ringdahl: Yeah, for sure. I think you make a really good point there. Your ability to execute and leverage AI will depend on a lot of the systems that you're using today. Spreadsheets, especially if different data is in different silos and different things, your resource management, your project management, and other things are in different places.

    To me, it is very, very hard to pull all that together and make use of it. What I would say, and sort of the way that I look at this playing out, is to look at this as an early phase. What can we do for the next 12 months? Some of the things we talked about were intelligent content generation, project planning, like just low-hanging fruit, proposal generation, being able to respond quickly, and just increasing the overall efficiency.

    We know right now that PS organizations aren't expanding, but how can you get better productivity out of your existing resources without actually asking to work more hours or longer or do double duty? So there's an immediate pick-up that could be had by leveraging some AI tools.

    The second phase of that is what I would call meaningful improvements that help efficiency and productivity, and some of that does require maybe the unification of some tools, because then that allows you to start to take advantage of some other things around, like resource management and skill planning, and start to bring those things together.

    Then maybe two years in and out, because I think trying to bite off too much at once is going to be a losing proposition. I think we are really building this in phases. So that last longer-term phase is really about automation, solution design, and larger and more impactful changes. These things may require long-term learning techniques.

    You may need a data set that is much larger than you have today to really get accuracy and better guidance on that data. So figure out what some of those long-term investments are and some of those long-term advantages, and put those in your later phase.

    Banoo Behboodi: Excellent. With that, I think I'm interested. I'm sure the audience or listeners are interested in hearing what strategies at Kantata are related to AI and how we're using AI to address our customers and just our technology.

    Ken Ringdahl: This is something I'm really excited about as a technologist—being able to bring some of these new cutting-edge features and functionality into our product. We've previously announced a new product that's not yet on the market but will be later this year, called Pulse. This is really coming back to that whole sort of client and employee engagement.

    This directly addresses those areas of our product vision. There are some things that we're looking at. Pulse will engage clients, understand sentiment, and really get that idea of, Hey, how is my project going? I am on budget, and I am on time. Is the customer pleased with how it's going? Are we satisfying their requirements and their confidence that we're on the right track?

    So I'm really getting those ideas because, as we've seen, a lot of that information comes back after the fact. The project is done, and you get some feedback that you wish you had known 2 weeks, 4 weeks, 6 weeks earlier, or much earlier in the project, because you can actually action it and adjust that.

    Where AI comes into this for us is that we have tons of historical data. We have thousands of customers who have done hundreds of thousands of projects. We know what good projects look like. We know what not-so-good projects look like. We know where things start to trend in the wrong direction and what causes those things.

    AI can supercharge that in terms of analyzing some of that data and noticing trends in existing projects based on sentiment and the progress of a project, which can really help our customers know before they realize it. Because oftentimes, a project manager will see a trend. But that trend doesn't happen until actual time has passed. And the idea is to actually pull that earlier and use AI capabilities to do that and to know very early and very quickly what may be happening and, maybe more importantly, what you can do to potentially improve and change something that's not on the right track.

    There are other things that we can do that we're exploring, and these are more future things. But from a client engagement perspective, you have Zoom calls, email communications back and forth, this document that you're working on together, and AI, again, back to generative AI, is really good at digesting content, delivering, and outputting something there.

    AI can do a really good job of analyzing all that information and coming up with even more information that can really inform a project's health, etc. A couple of other things I'll call out here are time sheets. This is something that we hear from users of our product all the time: project managers have a really hard time chasing consultants to enter their time.

    We're actually exploring some ways in which we can really remove a lot of the friction in that part of the product. And really, what we can do with the power of AI is, “Hey, we know what you are scheduled for. AI can look historically at what a particular consultant or an individual on a project has done.” And instead of asking them, “Hey, can you enter all your time for the last week or whatever it may be?” we could say, “Hey, look, we think this was the time that you spent this week with a high level of accuracy,” and it becomes more of a check like, “Yeah, that looks good.” Check. Submit it.

    It really, really reduces the burden on anyone to enter their time and improves that success rate. Resource management: we actually have functionality in our Kantata OX product today, which we call the Optimizer. That can really look at staffing and understanding; this is where spreadsheets really become super challenging: “Hey, I need a resource on this project.” In a spreadsheet, it's so hard to understand what the cascading effects are of that, and so we have this part of our product called the Optimizer, which really can give you scenario planning for resource management and uses AI capabilities and more machine learning capabilities to be able to propose to you an optimized way to manage your resources across many different projects.

    Skills are another area of understanding who has what skills. Matching those into certain projects and starting those projects. You mentioned earlier that it takes about 18 days to staff a project. Once the proposal is accepted, with the introduction of skills, you can better match skills to a project, which can increase the success rate of a project and, ultimately, increase the satisfaction of a customer and turn into repeat business. They're all interrelated. Ultimately, this results in a better outcome for the client and a better outcome for the PS organization as well.

    Then, maybe lastly, just financial forecasting. AI is very, very good at sort of doing financial forecasting, looking at your pipeline, your proposals, and work in flight, and being able to accurately understand what a financial forecast may look like.

    So those are a lot of the things I mentioned there. Some of them are in queue, and some of them are really ideas that are going to go into our long-term roadmap that we'd love to deliver to our customers.

    Banoo Behboodi: I love it. It's very exciting. With that, how are you or your engineering team within Kantata using AI to help make your everyday work more efficient and effective?

    Ken Ringdahl: We're sort of touting the benefits of AI and helping PS organizations. We are also looking to do the same. Our source code management platform is called GitHub. Many people are probably very familiar with it. It's the most popular source code management platform. GitHub has a product built into it called Copilot.

    Copilot is, as I mentioned earlier, a generative AI that is able to be almost like an assistant. Copilot is like an assistant to my developers. What Copilot can do is do some level of predictive coding. It can provide contextual hints. It can automatically generate some code. So, one of the things that's really important in delivering quality code is having really well-written tests.

    So you write some block of code or function of method that does something, and some things that Copilot can do are actually, “Okay, now go write me my unit test.” Again, in fact, it's also sometimes hard to get developers to write really good tests because they want to build that product, but just as important is building the tests associated with that. So these are some things that Copilot can do. And really, it's about increasing productivity and efficiency and really getting our developers, and I mentioned knowledge workers earlier, to focus more on those knowledge tasks as opposed to some of the remedial tasks.

    There's also IDE integration. An IDE is basically a developer’s environment. Think of an application where they're writing code every day. That application knows what code you're writing and can look at it. I think people have probably seen this. If you're using G-Suite and Gmail, Gmail now predicts what you're going to try to write, so think of an IDE integration that's very similar. They know the code that you're writing. And they can suggest what your next parts of the code will be. Again, efficiency and productivity gain overall.

    Then maybe lastly, some things that I think are really, really helpful are on the production operational side, just reams of logs, and as we've said, AI is really good at reading content and summarizing it. As far as analyzing logs and pulling out anomalies and abnormalities in those logs, that can really help us identify where maybe our products are doing something unexpected that we can dig into because that saves us a ton of time overall.

    Obviously, the engineers are going to be early adopters. They're going to really embrace this technology. But even outside of engineering, some things we're doing as an organization really have lots of revenue opportunities.

    One of the things we're leveraging is a forecasting tool that integrates with Salesforce, and I mentioned earlier some of our plans to potentially look at Zoom calls and emails as well. On the sale side, AI can be really, really helpful to look at and understand at what stage a particular deal may be. And it can also understand sentiment from a customer and really give a pretty accurate idea of, “Hey, this deal is really at this stage,” and predict, based on sentiment and tone in the email and other engagements, how the AI feels, that your success rate is going to be on a particular deal.

    And so we're leveraging some of that today. Again, we're not leaning on it completely. We're using it as a way of measuring our own accuracy, and it's just another input to allow us to be much more efficient as an organization.

    Banoo Behboodi: I would assume, in tangent with efficiency and operational effectiveness is quality. I would have to think that there's also a positive impact on quality with everything you just mentioned.

    Ken Ringdahl: Yeah, 100%. There is a positive impact on quality. I think giving everyone, like if I think of my developers, more time, but I will caution one thing on the quality side: this old adage of trust but verify. I think there's a little bit of that here as, “Hey, look, these AI capabilities and AI tools can really increase efficiency and productivity, but bringing it back to validating that and getting comfortable with those capabilities for sure is important.”

    Banoo Behboodi: I really appreciate the time you've spent with us, Ken. Amazing conversation. Thank you. As always, I've prepared you to know that I would like to ask a personal question. And with you, I'm going to do a double whammy. Ask you both for your book recommendations and also for a mentor who's been influential in your life and career to get you to where you are.

    Ken Ringdahl: I'll start with the book. It's probably about 15 years old. It's been read ever since then. It's called the Phoenix Project. As a CTO and a software engineering leader, I'm a huge proponent of DevOps principles. And CI/CD: constant development, constant deployment. This book, The Phoenix Project, is written by some industry experts on DevOps concepts and processes.

    I really recommend it because it is written almost like a novel. It's a story about sort of this fictitious auto parts retailer that's really fallen behind their competitors and really has to improve their systems. And it's told about how that retailer improves their IT systems in order to become competitive. And the most important takeaway from that book is about flow.

    I think of this all the time with my development teams: how do we get more efficient and productive? There are a lot of things we're talking about here today, and one of those things is optimizing work that goes through your teams.

    This is relevant not only for software development but for many other teams, and certainly for our PS organization customers and partners. It talks about how you optimize your flow and about being focused and working as a team.

    I highly recommend the Phoenix project. It's something that I've recommended to multiple teams in the past. When I recommended it here at Kantata, everyone sort of got behind it. They even did a book club to really talk about it and how they could take some of those concepts and bring them back into our day-to-day lives. So that's the book.

    As per my mentor, I've had lots of people in my career. They've been incredibly influential. I'm going to go a little bit off script in the sense that this is not someone who's been a sort of mentor of mine, but someone who I've kind of looked up to and who really was very forward-looking in the technology industry. It's Jeff Bezos.

    Everyone, I'm sure, knows Jeff Bezos is the founder and former CEO of Amazon. A lot of the things he did early on at Amazon really laid the foundation for modern microservices development today. So microservices development really broke things down into small, really small services deployed independently, and he did a couple of things.

    One is that he famously sent out to the entire company this API mandate, and in his API mandate, it was like seven different bullets. But he basically said that everything you build needs to be externalizable and have an API, and everyone must integrate with the API. No sort of backdoor? And what it does is force you to build standalone components.

    Coming back to the whole notion of flow, if you build standalone components, you can have teams that are very autonomous and can increase their flow. He also introduced the concept of a two-pizza team. What that means is that your team should never be larger than two pizzas. It means the team is too large, and it means there's inefficiency in that team.

    There are things that I have really adopted in my sort of roles as an engineering leader and driven consistency throughout my team because I strongly believe in the ability to be efficient and having teams that are just the right size to maximize that efficiency and productivity.

    He's someone who I've certainly looked at and modeled some of the things I've done after.

    Banoo Behboodi: Awesome! I love it. Thank you again for being here. I appreciate your leadership and vision at Kantata, and I appreciate you making time to be with us and doing this podcast.

    Ken Ringdahl: Thank you so much. I really enjoyed it.

    Brent: If you enjoyed this podcast, let us know by giving the show a five - star review on your favorite podcast platform and leaving a comment.If you haven't already subscribed to the show, you could do so anywhere you get podcasts on any podcast app. To learn more about the power of Kantata’s purpose-built technology, go to kantata.com. Thanks again for listening.