Treat AI Like a New Hire: Lera Kooper on Where to Actually Start

---

Luke Frye (00:00) Hi there, and welcome to Canopy’s Practice Success podcast. I’m Luke Frye, Accountant in Residence here at Canopy. And today I’m joined by a very special guest, Lera Kooper, for the first of a five-part series. I can’t wait to hear more about her and what she’s up to.

But before we dive in, I’m going to just have you introduce yourself.

---

Lera Kooper (00:33) Thank you. So, Lera Kooper, owner and COO of Accountability Services. We have a presence across the country. We have team members across the country, and we’re growing rather aggressively through M&A. So a lot of what we’re talking about today is directly related to my day to day in the business.

---

Luke Frye (00:58) I love that. It is a bit of a wild west today with all of the different technology and AI that’s out there. But because you have a background in supply chain technology, you’ve been able to think of AI as more than just a tool. It actually has a workflow component. And so that’s what I’m excited to talk to you a little bit about.

I would love to first go into just a real tactical question: what was the first AI tool you actually integrated into your firm’s workflow, and what problem were you trying to solve?

---

Lera Kooper (01:26) I think as we started exploring AI, or myself personally, you’re just testing the waters. Even the experts of AI have only been experts for so long. There’s still so much research and development going into it. And so for me, it started with just, how can I save time?

Probably refining my emails or my communications, or you’ll do voice to text, or have a conversation with Claude or ChatGPT and say, can you make this sound better? And that saves a lot of my time, saved a lot of my time.

I think AI meeting notes started saving a lot of time on discovery calls with prospective clients. I used to have an admin join me on the meeting to take notes so I could be completely focused on my prospect. That very quickly went away, because I have AI notes that I can then chat with and get nice summaries and recaps that I can share with the prospect or client, whoever I’m meeting with.

So a lot of it started with just exploring how I can save time.

---

Luke Frye (02:12) Sure. And did that admin go away, or did you reassign them to different, higher leverage work?

---

Lera Kooper (02:18) Reassign them to more higher leverage work. I think from the beginning we knew that is not the best use of somebody’s time, but that was the only option that was secure, or reliable, or, you know, fill in the blank available even at the time.

And so then they were able to refocus more on operations management and workflow and client experience, rather than just being a fly on the wall.

---

Luke Frye (02:47) Right. Lets them use their skills better. So in some ways, using AI and leveraging automation actually makes them more valuable and more human than they may have been before.

---

Lera Kooper (02:58) That’s the intent.

---

Luke Frye (03:01) Yeah. And I think in our industry, there’s at least a stereotype that we don’t want, like, we’re introverts, which I might be, but we don’t want to meet with people or talk to people, and we just want to be heads down in desks.

But I think really we want to build relationships and help people. And the more we’re able to get out of the data entry side of things, or just out of the boring work, and more into advisory in front of clients, the better we’re going to be at our jobs, the more fulfilling of the job it’ll be, and then obviously our clients will benefit. And so I think utilizing it to leverage our skills, it really makes everyone happier.

And I noticed the first thing you did — and I think, no, in part because of where AI is and where we know it works well — the first thing you did in your accounting practice was actually related to communication, right? And we often think of AI and accounting as some sort of automatic reconciliation tool, or automatic tax form filler, or something along those lines. But it’s actually more about doubling down on the relationship and communication.

---

Lera Kooper (04:05) Yeah. And I think it can help in both of those avenues. It can help with streamlining, improving, refining communication, or saving time on communication. It can also help with filling forms, data entry, filling tax forms, all those things.

Again, under the assumption that this is all secure, if it can do all those things, we’re able to now save time from the communication side of things as well as the technical side of things. And then just advise, help our clients act and move on.

---

[Recording break — trim in edit. Luke and Lera stop to check that Lera’s mic is live, confirm the audio is good, and agree the conversation is flowing. Recording restarts below and the timestamps reset to 00:00.]

---

[Recording restart — trim in edit. Luke recaps where they left off: note takers, and reassigning an admin to more operational tasks, on the idea of highest and best use. Question re-asked below.]

---

Luke Frye (00:20) I’m curious how you decide which tasks are good candidates for AI versus which still need a human.

---

Lera Kooper (00:30) That’s a good question. I’m sure this will lead into later conversations, but I compare it to a new hire. I’m not going to give a new hire full authority within the first week on how to handle client escalations, or how to handle billing situations, or even advisory if they don’t know how we operate. And so similarly with any new tech, with AI, I’m going to start with the tasks that require the least professional judgment.

So then it’s very easy to define what good looks like. I can tell Copilot, I can tell Claude, or whatever tool, I can say, this is what I’m looking for. And then either it worked, or I need to tweak some things, or it didn’t work. And then you continue to teach it what you’re looking for. And then obviously the longer of history you have, the more professional judgment can be mirrored based on what you’ve taught it. That’s kind of how I compare it.

So I would say anything like communications, shortcuts, data entry, bookkeeping, reconciling. We’ve worked with Basis for a few years on some of our clients, and especially the high volume clients. And they have relationships with banks, so they can pull statements. They can save a copy of the bank statement to our SharePoint for us, so then we don’t have to do that. And then they do the data entry and reconciliation, and then even have levels of confidence, which is where we’re starting to see it triage professional judgment through its own work, but then still flag it for us to review and say, yep, where it was confident, I’m confident, and then these are the opportunities where we continue to refine that.

---

Luke Frye (02:05) Got it. So it sounds like, especially early days, keeping a human in the loop, and probably more human time in the loop. And then once you have enough confidence that it’s able to execute on a certain type of task, or project maybe even, then you will let go of the reins a little bit.

---

Lera Kooper (02:26) Yeah. And like any team member, you give them a measure of authority, or expectations, or KPIs, and then you have weekly check-ins on, is the response rate actually one business day? If it is, if it isn’t, you troubleshoot, and then you move on.

And so I think technology, AI, can be the wild west, but you can reign it in a little bit with some oversight. And then the more defined your processes are, the less oversight is needed.

---

Luke Frye (02:58) Right. Which probably is two sides of the same coin, on having really standard process and workflows. It helps both the people onboard and get up to speed faster, as well as any new tech or AI agent that you might work with.

---

Lera Kooper (03:12) One hundred percent.

---

Luke Frye (03:15) Out of curiosity, how do you typically define your SOPs?

---

Lera Kooper (03:19) How do I define them?

---

Luke Frye (03:21) Either tools, or methods, or…

---

Lera Kooper (03:24) So my biggest project over the course of the last year was documenting SOPs. Anything I do, anything my team does, if they do it under the definition of good, I want them documenting it.

So right now, I know it’s not the best tool, but it’s a good way to just collaborate and create a library dump, is OneNote. And so I have a master SOP with sections by department: how we do things, how we onboard clients, how we do a technical review of a tax return, everything. And then even if it’s pretty bare bones, it’s in there.

We’ve kind of gotten into, or I’ve gotten into, working with Copilot agents. That’s something I’m exploring right now, because what I would like to do is take the content that’s there and turn it into basically a chatbot within our Microsoft tenant. So any new hire can say, how do I onboard a new client? Here’s the documentation, here’s the steps.

And so then we create our own secure database of information on how we do things, not on what the internet says is the average.

---

Luke Frye (04:38) Exactly. I love that.

Okay, so this one does require a level of vulnerability. Has any AI experiment failed or got pulled back? And if so, what did you learn?

---

Lera Kooper (04:51) Yes. Failure, I think is a strong word in this sense. I think we didn’t give it the resources we needed to govern the project. So it wasn’t a failure in terms of we lost clients or we lost a lot of money. I think it’s just something we could have done well if we did it right. And then we didn’t dedicate resources to it, so then it just kind of got put on pause.

And what that was is our use of a tool like Basis, where we wanted to just, in one sweeping motion, put every single TAS client on there, but we didn’t have enough governance, or like one champion that says, this is the way that 90% of clients should be templated in Basis. And so then every single accountant ends up using it a little bit differently. And then you don’t really get the efficiency gains that you went in for in the first place.

So, love the tool, love what the tool does, we still have a handful of clients on it. That one accountant of ours, that’s her client group. But we have been so busy with other things that we haven’t had the resources to go in and define good in our use of that type of tool. Because then it does just turn into the Wild West and inefficient. And when you’re onboarding 50 team members, who’s training them?

That’s why it’s been put on pause. And like I said, failure is a strong word, but that’s one that I wish we would have approached a little more thoughtfully.

---

Luke Frye (06:50) And for people who may be less familiar with Basis and tools like that, could you just give us an explanation of what those types of AI tools do?

---

Lera Kooper (07:00) Yeah. And I think they’re getting funding, they’re constantly developing, so what I say now is probably two years old and will be out of date in six weeks.

How we utilize them is for the data entry, like auto AI data entry. They pull bank statements for those clients. They reconcile the accounts within Basis still. They save the statements to our SharePoint, where we have a lot of our long-term storage. And then they’ll flag, you know, with levels of confidence, what the data entry was.

You can create formulas for, you know, payroll journal entries. Because we know when those come in, it’s wages and taxes, but really taxes, half of it is wages. And so you need to make adjusting journal entries. So you can teach it all these things. It happens automatically.

And then, for our team, which I think is really the advisory win, is it then creates these summaries of risk areas, or opportunities, or here’s the highlight. So I think it’s a really good tool for accountants that already have professional judgment, to just do more and have to spend less time isolating trends on their own and all of that. So it’s a shortcut.

---

Luke Frye (08:50) So this really is an area where the AI hype is saying we could replace a bookkeeper on some level.

---

Lera Kooper (08:58) I think, yes, that could be maybe a fear-based statement, that it could replace a bookkeeper. It could also train a bookkeeper to advise in how to isolate trends quickly.

It’s like, if I look at a balance sheet, I could immediately tell you, these are the three red flags. This needs to be cleaned up before it goes to tax. A new bookkeeper may not be able to do that. But if they’re working with a tool, and I think Intuit is doing this now, lots of companies do this, it’ll flag, you know, this is weird, this is a debit when it should be a credit. And it’ll just help people, I think, recognize anomalies that maybe they would have received from their senior manager before.

I think with the compression in the industry, there probably isn’t a lot of time spent on mentorship as there used to be. So I think for people to actually want to develop, it can be a good tool. But again, there is the danger of professional judgment, is you want to trust but verify the product.

---

Luke Frye (09:18) Yeah, professional skepticism. That’s what we’re trained for.

Well, that segues in pretty nicely to just how your team has responded to working with AI. So you used the example of your admin being in, taking notes, probably not the most fun job. But then there’s also this idea of highest and best use, and how we’re using technology alongside our team. And then I love what you mentioned about giving capacity for managers to provide mentorship.

So just curious how that’s been playing out so far.

---

Lera Kooper (09:52) It’s been going well. I think everything takes time, and as exciting as it is, rolling anything out takes time.

But our team is a lot more client facing now. I would say our bookkeepers and accountant level team members are so much more client facing now than three years ago. And that’s because there is the additional time for mentorship. There’s the additional, I think, just resources and angles that you can look at things with, due to technology.

We do still have one client that sends us scanned images of his paper ledger. Ten years ago, I used to go to his house. Him and his wife own a couple of rentals. Still sends that. So some movement is slower than others.

But I think within the team, it creates capacity to just get through the almost formulaic work faster, so then they can interpret the results and personalize it to the client, spend more time in that zone of work.

---

Luke Frye (10:50) Focusing on relationships.

---

Lera Kooper (10:52) Yeah. And I think that’s what’s going to be the biggest change in the next handful of years. Everyone wants the relationship, but I think now we’re going to have the capacity to do it.

---

Luke Frye (11:10) Yeah, we’ve been talking about advisory forever, and I don’t know that we really had the capacity to do that.

---

Lera Kooper (11:18) Yeah. And I think everyone defines it a little bit differently. Some might define it as a tax planning meeting. Some might define it as a one-on-one session teaching your client how to read their balance sheet. It’s all advisory.

---

Luke Frye (11:38) Right.

---

Lera Kooper (11:40) It’s not this one time productized service. It’s just this layer over how you operate, how you interact with your clients, if you’re interpreting the data for them.

---

Luke Frye (11:52) Sort of a mentality.

---

Lera Kooper (11:54) Yeah.

---

Luke Frye (11:56) Awesome. I love that.

We’re going to switch over to predictions now. So, I’m just curious, which accounting tasks do you think will be fully automated first? And what does that free accountants up to do, actually? I know we talked a little bit about that. Anything you think you would add?

---

Lera Kooper (12:15) I think we did cover that, and I think we’re almost there. Honestly, it’s just data entry and reconciliations.

The big one for me that I hope comes soon is relationships between the services, or the companies that do that work, and banks. So then if they have relationships with all banks, then you can get statements from all banks reconciled and saved to your Canopy or your SharePoint or whatever. Whereas I think a lot of companies have relationships with 30% of banks or 40% of banks. So I think closing that gap is what’s going to really send this into a fully automated process.

---

Luke Frye (13:00) Yeah, interesting value prop, banking being one of the biggest pain points there. And part of that may have to do with regulation, but then also partnerships and different neobanks that are available out there.

---

Lera Kooper (13:15) Banking, and even companies like [inaudible], any form that needs to be reconciled against the books, whether it’s a statement or a paycheck. I think the more those relationships exist in the network, the less human touch will be required to just get the data right.

---

Luke Frye (13:35) So, you know, I think a lot of people are afraid of AI, right? And so we touched on it a little bit, but a lot of what we’re saying, and what I believe, and what Canopy believes, is leveraging technology is a way to really empower people to do higher level work.

Do you see a tipping point at which we don’t even need to be in the books or working with a client, they can just chat with an AI software instead of us?

---

Lera Kooper (14:05) That’s a good question. I think there will always be a need for human relationships, and so there will always be a segment of people that want that. You have people that are price shopping, maybe, and that’s why they go to TurboTax versus working with a firm that does have that relationship. And so I’m sure that option will still exist.

When everything can be fully automated, there probably will be a segment of people that want to go with the budget route, because either their situation is simple enough, or they’re willing to spend their own time figuring it out instead of spending it on their business or whatever activities they have going on. So I think that option will always exist. It’ll just be a better option as time goes on.

But I think people will always want relationships. And that’s why the focus is now to lean into technology as much as you can, so then you can be the best at providing that human experience when the time comes where you had to be one or the other.

---

Luke Frye (15:10) Interesting. So in five years, would you say that most of your work will be just relationship management, or?

---

Lera Kooper (15:18) I think so, yeah.

---

Luke Frye (15:20) I love it.

Will the firms that wait to adopt AI be able to catch up? So I think a lot of people feel really left behind already, but conservatively making sure that data is secure and risks are mitigated. Do you think that’s rational, or is there a point of no return? Have we passed it?

---

Lera Kooper (15:40) Another good question. I feel like we’re behind with AI, honestly, because there are so many new options, and you don’t want to dilute your focus. My job is not to do R&D for a tech company, and so I still have a job. So then the fear is going in the wrong direction, I think. And that’s, again, assuming whatever candidates we’re considering is secure within the tenant, information data is good.

And I think there will always be room to, I don’t think anybody will ever be cut off or be sunset from the industry because they’re not utilizing AI. I do think it will have a significant impact on their lifestyle, because then it’s just a race to the bottom on price.

So if you’re looking to grow your company or grow a team, then you have to utilize the tools at your disposal. If you’re happy being a one-man shop with a certain set of revenue or income, your life could be much easier and you could probably work a lot fewer hours if you utilize technology. But could you still get a paycheck? Probably. I think it’ll exist. It just depends on what your goals are and what you want your life to look like.

---

Luke Frye (17:05) I love it. I love it. And, you know, rightfully so. Some people may just prefer a different way of doing something that’s maybe less expensive, but less automated.

---

Lera Kooper (17:15) Yeah. And I think, again, some clients really want that human experience. They still want to sit down with their CPA and watch them prepare their return. We don’t do that, but we do have some elderly individuals that they mail in their documents and we scan it into Canopy for them. And then from that point on, for the rest of the team, it’s like normal processes, because it’s scanned in, documents are there.

So there will always be a segment of clients that wants that.

---

Luke Frye (17:45) I love it. Okay, this is great.

One last prediction. Do you think AI will eventually make it easier for small firms to compete with large ones, or will it actually widen the gap and make it harder?

---

Lera Kooper (17:54) Large firms typically have a much larger budget, so then they can explore a lot of options and probably figure out what the best is, and then spend time rolling it out. Smaller firms don’t have the same budget, but they can typically move faster. So if they find something that works, they can typically implement it throughout the team pretty quickly.

So I think there’s a lot of opportunity for small firms to catch up to big firms in terms of efficiency and profitability. Going from like a sub $10 million firm to a top 100 firm, that gap has to be bridged by more than just tech. That’s leadership, that’s team members, that’s a lot more than just technology. But I think if we’re looking at business health, they could probably catch up.

---

Luke Frye (18:50) I love that. I love that. Awesome.

Well, that was topic number one of our five-part mini series with Lera Kooper. We’ll be back for more with a handful of other topics, so be sure to tune in.