Learn the “one-year rule” for evaluating new AI technologies. Discover why 50% of AI startups will likely disappear within two years. Roman explains how to identify proven solutions versus experiments.
For firms focused on staying investor-ready and scaling with confidence, this matters. Understanding which innovations to implement versus which to monitor is critical to reducing risk while accelerating growth.
Watch the full series to discover how to accelerate growth and scale with confidence.
Learn more about solutions for investor-backed and high-growth firms at https://www.rightworks.com/solutions/pe-backed-firms/
Speakers:
Derek Distin, VP of Community at Rightworks
Roman Kepczyk, CPA.CITP, CGMA, PAFM, Director of Firm Technology Strategy at Rightworks
Transcript
Derek: Am I looking at dumb or smart AI coming into the industry right now?
Roman: Well, when we talk about artificial intelligence, it is narrow AI, which is built on what’s called generative AI solutions. Those are the things that we’re using, which is like Google Gemini, it’s search on steroids, ChatGPT. Basically, you ask it a question, you make a request, and it gives you an answer that is authoritative, or at least it sounds authoritative, that you requested out there. And so we see firms, marketing departments actively using generative AI that’s out there.
Other variations of AI that we hear a lot about is the automation components. And so when you see about tools in the practice management space, the workflow space, the email space that does auto respondents, it uses the combination of that generative AI, big database, large language model to give you an answer and a script or a recommendation for a process, which is more akin to machine learning or robotic process automation that’s out there.
Now, this is all going to change in the next two or three years as agentic AI becomes more prevalent in the accounting profession. And what agentic AI is, is agentic AI has the capability of going beyond that large language model and going out to a secondary resource of information and then running those processes on that and bringing it in. So it’s actually expanding to multiple databases.
Derek: There’s a hype train right now that AI is the solution to everything and every product seems to run its own hype train. Every AI product seems to run its own hype train. Do I need to be jumping onto every single hype train that comes by?
Roman: No, actually, Derek, we would encourage you to avoid the hype train, you know, be aware of it. But again, talk to your peers that have been successfully using these different tools before you implement it.
Derek: How long should I wait before I look at implementation of tool A or tool B?
Roman: You know, for me, when we talk about successful implementation, the standard, the hallmark is if someone has used it for a full year rolled over and paid for it, then they’re probably a valid reference.
Derek: How many of these AI companies, these startups are going to succeed? How many are going to fail? How do I know which ones are going to fail? And how long should I wait before adopting a brand new AI technology?
Roman: Okay, when we talk of the boom of the bust and how to identify technologies, I would look at someone who has been working to solve your problem for probably two or three years, and they’ve had a solution in place during that time that has constantly evolved and improved and is being utilized by an increasing number of your peers.
But when we talk about the hype cycle, someone that just was created last year, I wouldn’t. I would listen and learn, but what we see is if someone has a good idea, it’s going to be copied by the more established players or even acquired by those established players.
I would guess that over the next two years that 50% of the AI new startups that are out there will have disappeared completely, and any intellectual property that they had was valuable will be gobbled up at, you know, cents on the dollar by these other providers.
I think the number of successful AI ventures is rolled out now, if it is north of 50%, I will be completely surprised. I would anticipate that their burn rates, most of them if they don’t get acquired, will probably disappear over, like I said, the next two years that are out there. We saw just a gold rush of money coming in and now they got to deliver on that. And unless accountants start using it, adopting it, they’re going to be gobbled up or disappear.
Derek: And what should I do if I’m looking to evaluate one of these technologies?
Roman: Well, if you’re evaluating a one-off technology that solves a specific purpose inside of your organization, we’ve got to do a basic ROI calculation that if you invest in this tool, how much time is it going to save? How much time are you actually going to save by implementing it.