Perfecting Processes with Isaac Perdomo

In a previous episode of the Big Four Transparency podcast, we talked about how much money some accounting firms are saving already by moving away from credit card processors and the fees that those come with and moving all of their payments from clients towards ACH and other forms of direct bank transfers. Now I've worked at a startup that didn't have any kind of tracking or any kind of a real solution around that and it made accounts receivable and accounts payable an absolute nightmare. But look no further because the sponsor of today's podcast is Forwardly. I met the awesome people of Forwardly over at Bridging the Gap conference, did a demo of their product. I am disappointed because personally I can't work with them unfortunately because of some restrictions on Canadian banking. But for any US listeners who are looking for a new solution for your accounting firm, Forwardly is a great solution for you to handle all of your payments, AR, AP, as well as a great approval workflow. So check them out. They're linked in the podcast description. Thank you very much. Hello and welcome to the Big Four Transparency podcast. I'm joined today by Isaac Perdomo, the co-founder of OPSR, an agency specializing in workflow improvement and automation, mostly for accounting firms. Thanks for joining me on the pod, Isaac. Yeah, appreciate you having me, Dominic. Yeah, my pleasure. I have seen you all around recently. You were on Automation Town with Chad Davis, who recently was a guest here. And so that kind of brings us sort of into the first topic of you're kind of a technical background person. You studied in engineering, specialized in automation and kind of that suite of tools. How did you end up sort of in the accounting firm niche? Yeah, that's a great question. Multiple answers to it. But one of them is that I've always been interested in money and accounting and finance. And I've been that kid that, you know, tracks his income and expenses all the time and still do to this day. My dad was an accountant and CPA as well. And so that played a role. But ultimately, we've been doing this work for about four or five years and we had sufficient data on, you know, what are the industries we did better with? So we did an analysis about two years ago, as well as, okay, here are clients, the different industries, who do we enjoy working with? Some industries that we were able to work with, like a couple of agencies, a couple of e-commerce companies, we just didn't have good rapport with for whatever reason. And ultimately, accountants are nice people for the most part. So we really enjoy working with them. They also had a lot of opportunities for automation, just because of the nature of their work. And we said, why not? Interesting. Yeah. I found, I mean, I also kind of am selling into accounting firms. I have found it to be an industry where people are a little bit slow in general to adopt, to change. Yeah. I'm curious if that's something you've encountered a lot too, or how you've kind of dealt with that. Yeah. I think we're in a position to capitalize on that in a way. Because the fact that they are sometimes slow to adopt, I would assume it's partially because of their capacity to do it, their bandwidth, and sometimes there's technical skills. So I see that as a positive for us, honestly, because then we can, if they're interested, assuming their interest is there, which usually is, if there's a payoff, then we can support them in making that transition. Yeah. Yeah. That is the flip side is I will say, I believe most people kind of making decisions in firms are highly rational people. And because the industry for so long, now it's maybe decoupling a little bit, but for so long was so based on the billable hour, people really understand the ROI of cutting how many hours you need to put into something, right? Plus when you couple that with the fact that there's such a high demand right now and shortage of supply, in general, being able to do things more effectively will kind of mean more revenue for the firm. And I think people are able to kind of take that mental step quite well, right? Because of the sort of financial education background. So yeah, definitely. Yeah, that's really cool. So what was like, what would like the typical project look like for you when you're going into these firms? And I'm sure there's a ton of variability, but like, just to give us a general sense. Yeah. Yeah. So effectively, we walk through a process and we're kind of refining this internally of auditing, streamlining and integrating their systems. And really looking at it a little bit more holistically. So looking at the people, the tools and the process to assess, you know, if it's worth engaging in the first place and what they have, they can automate. So you can envision, and we have this kind of like mapped out in our internal documents, it's like a nine step process. In the assess kind of initial phase, it's really about benchmarking their capacity, assessing, you know, how much what's the upside really of, of, of efficiency and automation. I could dive a little deeper into that later if you want. But then it's really about clarifying their processes. Do they have any documentation at the moment? And if they don't, usually it's kind of interviews with different stakeholders to assess how things are currently getting done. And then from there, also, we kind of take a holistic look at their tech stack to see if we can find either gaps or opportunities to consolidate certain applications in that in that in their tech stack. From there, we can optimize how them implement or integrate some of their systems, depending on what we found in that initial kind of auditing process. And then of course, we kind of automate and make sure that the team adopts the new tools and processes, and then kind of rinse and repeat from there. Okay. And in that process, like, how often are you finding, for example, like, two softwares where potentially you could just match them up using make or Zapier or some like direct API, but like the people in the firm are instead downloading a report from one and then just basically like manually and putting it into the other piece. Like how often is that happening? All the time. I like to say that there's just almost no firm that doesn't have any opportunity for an automation. So it's almost nine times out of 10, we find a ton of things that could benefit from that their people are doing manually at the moment that they could, you know, delegate to a robot, essentially. Yeah, yeah, I find that interesting, like in my personal, like learning journey through university to like become a CPA, we had accounting information systems. And my my kind of complaint with a lot of these courses is they're not very practical. Like I think that it should have really been done as like, pretend you're managing a firm, here's something, take it through a workflow, and you need to like understand how everything connects. But instead, it was like this very like high level theoretical, like, here's what an ERP is, blah, blah, blah. And so I completely disconnected from this class, I was like, I think this is useless. This is dumb. And I just I completely disconnected. And that's one of those where like, I'm on the other side now. And I'm like, Oh, man, that was like, probably top four most important classes, right? Like, and again, like, I think education wise, this should have been done in a more practical manner. But at the same time, like, the the value of it, like was not being appreciated by myself and a lot of other students. And I think it's not until you've had that practical experience somewhere that does not have these systems in place, or has like a poor tech stack, that you actually like start to realize like the the absolute just power of of having that, you know, buttoned up properly. Yeah, definitely. That's something that I think it's one of the challenges of, you know, just traditional education, they have to be broad enough to help everybody, but that they can be specific enough, because there's a limit of, of time. But definitely, I think that ultimately, the tools we use evolve, we evolve with the tools we use. So there's always a kind of like evolution that needs to happen for us to further improve the way we work. And our tools are just a big part of that. Yeah, yeah. So okay, so one of the big pieces is kind of streamlining the transfer of data from kind of one system into the next, making sure that that kind of flows properly. Is a lot of this as well? Like, are you are you like building a lot of dashboards internally? Like, what are people usually looking at, or looking to kind of get out of this beyond like the proper integration of different tools? Yeah, that's a great question. And it's funny that you mentioned that because we're kind of in the process of kind of expanding some of our offerings. And that's where we'll be kind of like a key part of that, specifically something that we did with actually an automation town and kind of we build it live is a kind of effective hourly rate report. That's something that we're now going to start to at least suggest to everyone we work with, for various reasons. Actually, I feel like that's the metric that you can most directly tie to revenue and you can impact with automation. And I can give you a quick example of, of what that would look like. Let's just say you have a bookkeeping firm, and they're charging, you know, $2,000 a month for their particular package. And let's say it takes them 20 hours to deliver on that. Most firm these days, at least the model firms we're working with are often not charging by the hour, there have some sort of, you know, monthly package. And there's some arguments there's people kind of want to just not track their time, but I think there's value in doing it. In any case, you have here $100 an hour $2,000 a month, 20 hours to deliver. Of course, if you are able to charge more, you increase your prices, you know, better marketing, better branding, that kind of outside of the scope of what we do. But if you increase your price to say $3,000 a month, now you're making $150 an hour. But also, another thing that people can sometimes fail to realize is if you lower the hours that it takes you to fulfill that particular work, you also increase that number. So if you go say from 20 hours to 15, now you're making $200 an hour for that particular client. So through a improvement on pricing and efficiency, that particular example, you're doubling the revenue that you can support in your firm. And that's how we see, I guess, our role in helping our clients be more successful. Because at any given time, you have different levers that you can pull, right, you can pull pricing, but you don't want to be so expensive that people don't buy. So you kind of have to balance pricing with efficiency as well. You know, to balance that out. Which is funny, because I feel like those are probably like the two biggest things like holding the industry back, right? Like, a lot of independent firm owners like notoriously bad at pricing, because like, we're not generally, you know, conflict. We're generally pretty conflict averse people. And a lot of people like they just they like their clients, like you, you gain an affinity for your client, and you're like, ah, like, I don't want to increase pricing on this person. Like, I see their business, it's not going that well, or, you know, whatever that might be. And so I think pricing becomes a sticking point. But then at the same time, a lot of people are kind of resistant to change, and like trying out new things and new systems. And eventually, at least one of those two things needs to break, ideally both. Because otherwise, what you're going to do is you're just going to kind of run yourself into the ground by working outrageous hours to make everything work and for me to make the business model work financially. And I think that's a lot of kind of what's gotten the industry into a hard spot is that that is kind of the case too often, right? So I love to see people kind of promoting on both sides, like improving efficiency of operations, and then at the same time, the need to actually like address and look into pricing. So I think what you do is incredibly important for the industry, honestly. Yeah, I appreciate it. Yeah, my pleasure. And so for some of these projects, where people, like you mentioned, are kind of looking at dashboards, what are kind of some of the other metrics that they're that they're typically tracking? And that are kind of driving that? Yeah. It will, it will, of course, depend on the firm. But as it relates to particularly the efficiency of the firm, we'd like to look at the effective hourly rate. So effectively, you know, the revenue you're capturing for a client, and then how much time you are spending to deliver on that. Something else we also like to look at, if available, if we have time tracking, we can make that happen. It's the utilization of the team. So simple math of saying, hey, this how many hours does the particular team work? And how much were they supposed to work? And running that running that calculation? As I like to say, like, I don't think there's a limit to how many, how much? Let me put it this way. I think you would want to aim for as high of a hourly rate as you can. But I think it's dangerous to say the same thing about utilization, because it's not you don't want to work your team, you know, to their fullest extent. So there's a balance there, of course, of trying to have that be within a healthy range. And then utilization for what it's worth can be dealt with in multiple ways, right? Like there's increasing actual hours worked, but like, you can also really hammer down on admin time. I haven't seen this in accounting, but I've heard of like some tech companies where they're like, they'll give every employee a $5,000 Upwork budget, or like something like that. And they're like, so long as you're not like throwing confidential data into this, like anything that you think you can kind of take off of your own plate and outsource, right? Or having a higher percentage of like admin staff will typically kind of be a lower salary than kind of a lot of the kind of practitioners and having a higher percentage of admin staff. And then you can then offload more of the kind of administrative work. And like a lot of these people are super competent. Like I know some people who are admins back when I was at Deloitte, and they're now like, they're not necessarily fully certified kind of in tax and whatnot, but they've found pieces of actual client work that these people can now take on. So they get to expand the scope of their duties, they get more variety in their work, and they're probably more engaged. And again, you've now pushed the dial of like, what your kind of tax manager is looking at has now kind of moved up to just sort of like higher and higher leverage tasks. And that's like a way again, like utilization can kind of be improved, where it's like you're not spending any time on kind of these non-administrative tasks. Yeah, that example of the Upwork budget, like I think is like, just brilliant. That's insane. I hadn't heard that. But it's funny, because actually, that's also an option that we discussed. For example, let's say we're doing process mapping sessions with a client. And we say, actually, this step, maybe there's a lot of variance, maybe we don't want to automate, could we at least delegate it to a administrative person or to a less expensive resource, a less expensive human being, so that still, even if there's people doing that particular thing, it's still a kind of, you know, you're lowering the cost of that deliverable, so to speak. Yeah. And it's funny, like, this is not meant to at all be like, like an elitist thing where the person's like, I'm too good for this task. It really truly is that like the person themselves, there's like a learning curve there, and they need to learn what they should be focused on what they shouldn't be like, I had Wade Runge on here, who was a partner at a very small firm that got acquired by a much bigger firm. And he was talking about how like his billing rate, like exploded, right? Where they were like, okay, now like, as a as a director here, we expect you to bill out so much per hour. And he was like, Well, I wanted my clients to receive the same quality of service. So what he had to learn to do was to leverage his time and to be able to, again, stop spending time on these, like, tasks that didn't need to be done by him, and consider the billing rate of each kind of level of person and then assign accordingly, right? So you almost become like a resource manager at a certain level when you're trying to, like, make your budgets work and all that. And I find it really interesting. And it's, it's very strategic. And it's not a case of like, Oh, I shouldn't be I shouldn't have to deal with this, right? It's truly like a strategic decision. It's not like an ego driven one. So yeah, yeah. And I think there's also a lot to say about, hey, you can do it with the best, your best interest at heart of, hey, there's another human being that will have now, you know, more work to do and delegation can come from a from a good place. I agree. Yeah. Yeah, absolutely. So for someone who's like trying to get started in this, like, if you're just like, you know, you're, you're like an analyst level person, and you say, like, I want to be able to champion this internally. Where would you recommend they kind of like start looking? What size of firm that might change the answer? I like that. That tells me that you've been through this a couple times. Let's say like a medium to small firm, because I think big firms are a lot harder to drive impact in as a person. Actually, I think that regardless of, of size, to an extent, you want to kind of get buying first from your stakeholders involved. And that's usually done trying to establish what's the payoff and the return on whatever it is that you're about to embark on. And this is going to partially kind of the first step we often recommend, I mentioned it earlier, if like benchmarking your capacity, it's saying, okay, how many team members do we have? How many hours are they spending on client work? How much are we billing? If we are fixed fee? Are we actually making the amount of, you know, the effective hourly rate we would want? And can I make a case that, hey, if we, you know, we're 25% more efficient, as we saw earlier, we would be able to support a higher revenue, we would be a more profitable firm. And usually having those initial conversations is like pretty key to get people on board. Because if people are not on board, then you know, nothing's gonna happen, nothing's gonna move. So that's really where I would start. And the second thing that comes to mind is just finding use cases. Because it's all a well to establish, hey, this is the impact we could have if we save this amount of time, but how, how exactly that's going to be usually the next question. So try to come up with use cases that you believe based on, you know, the operation of your firm, you will benefit from which can start with what you do as a contributor. That could be typically contributors are more in touch with what it is that they're working on a daily basis. So they'll have more insight to assess what could be automated or not that, you know, they the higher level managers. Yeah. And then if they want to learn how to like do the actual automation, like, I think that's a little bit of a daunting thing. Like I've, I've learned some myself, through kind of necessity, right? Like I changed email providers, and that didn't integrate properly with the no code tool that I built big for transparency on. And so it was like, okay, I guess I'm going to learn how to do webhooks, right? And it seems so intimidating, but it was actually super easy to do. And so like, for people who are looking to kind of overcome that first hurdle, like, what do you think is like a good? I mean, first of all, level of expectation to have and be like, what what kind of tools should they be looking at? Certainly, different tools will have different levels of ease of use. SAP here is one that's often cited because it's tends to be very beginner friendly. And you can just get started with simple automations. So definitely, that would be a good starting point. I'll also suggest that now we all have the greatest and latest resource available at our fingertips with chat GPT. And very often is there's a particular thing you want to learn, you can have that resource as YouTube as well. We have some videos on some of those topics like webhooks and other technical tutorials. So that could be a good resource as well. I've also find a lot of value in communities, because then you're also learning what others are doing. So, you know, chainless block to chat Davis and the other people that have communities in the space, but that's tends to be very helpful as well. For them to start getting an idea of, of, at least the how the what specifically could be done. Yeah, I'm really happy those communities exist. Because again, like I said, it seems so daunting. And and people have this kind of like self limiting belief that, you know, people have this kind of like self limiting belief of like, well, that's not what I do, though. Like, I'm, I'm a CPA, like, that's not what I do. And I think once you get over that and build your first automation, and can actually see like, hey, like, this is going to save me. And, you know, a lot of the a lot of the early ones, the bar doesn't need to be that high, but this is going to save me 2020 minutes a week. And it took me an hour to build like, it's a pretty quick payoff, right? Yeah. Um, plus, again, you're you're kind of advancing your own technical knowledge, because, you know, when you're at a small firm, and they embrace you doing this, I think it's such a good opportunity for you. Because regardless of what your future ambitions are, if you end up at like just some other firm, or if you end up like out on your own, like this is really something you can do to stand out. And I think kind of the next generation of accountants like is going to be a lot about a lot more than just doing the technical work, right? I think increasingly, as we become supported by technology, being able to kind of harness those properly matters a lot. Or on the flip side, there's other things like being a relationship person, or, you know, whatever, right? But I think there are a lot of things that you can do to distinguish yourself. And one of them would be, yeah, being that really good process driven person who can kind of like continually improve the actual business itself. So yeah, and I find that work incredibly interesting. Like I really liked it when I was at Deloitte, like anytime I got an opportunity to do that. And obviously, it's kind of different at a firm that big, but it was like, Oh, I got to build like an FX tracking template that like got used pretty widely. And it was like, Oh, that was that was fun. Right? Interesting. That's funny. I wasn't expecting you to have built automations in Deloitte. I love to learn about that. Yeah, they weren't really like automations per se. Like it was it was really just like, there was a lot of like manual tracking of like, average cost basis on like these FX transactions. And there wasn't really kind of any resource. So I like I built this thing and actually got like, shared out pretty widely. Which was which was kind of cool. Because I was like, again, I put a lot of effort into doing it once and putting the time into doing it where I was like, I'm gonna I have eight other clients I can use this for. So I'm just going to use this again. But it wasn't. Yeah, not not actually building like for real automations for it, unfortunately. But that's I think that ultimately, that's also part of it, right? Like, it's just finding this sometimes small use cases. Even like, you know, rules on your inbox or little macros on Excel that can save you time, I would still categorize that as automation. Even if it doesn't include like, you know, a much more modern automation tool. Yeah. And for what it's worth, like for CPAs, or just accounting professionals who are listening who like, think, you know, maybe this is just like a detour, like, the highest paid job I've ever been offered was as a senior process improvement leader. And like, I've only done like little things here and there to like, but I have a very keen interest in improving processes within a company. I have never been offered a job for nearly that much money, like in my entire life. And anyways, it didn't, it didn't kind of work out timing wise and all that. But like, there's, there's, are you able to share it? Because you got us curious now? Yeah, honestly, like, I don't mind. It was, I think, like the total comp, like, I was making like, just over 100,000. And the total comp package was like 185 or $190,000 and it was like it would have been for like an insurance company. And again, they were like, we love hiring CPAs for this because you really like, tend to have that kind of very logical approach to things. And then kind of at the same time, like they so they so they like the the thinking pattern that a lot of us kind of have. And at the same time, they were like the the five or six things that you can point to like tell us that you're going to be able to figure this out. And I was like, wow, that's that's quite the ROI on like a few small improvements I've made throughout my careers. And I think learning these tools can like really, really, really pay off. So yeah. Yeah, definitely. Interesting story. Yeah, it's funny. I'm happy you asked for like the deets because I'm not, you know, I'm on the channel preaching salary transparency. So I should yeah, I was gonna mention that too, like we want transparency here. Yeah. So thanks for the challenge of that. That's perfect. Um, so okay, so we talked about AI a little bit. That's obviously like the elephant in every room. And so how do you think that's gonna transform or has already transformed kind of like what you do and kind of processes internally? And like, I would love to hear something like as specific as possible, like in terms of like an application of this. Yeah. I think that fundamentally, what AI has allowed us to do is at least looking at it from the lens of like automation, it, it has allowed us to go beyond rules, right? So traditionally, a lot of algorithms, a lot of softwares, pretty rules based, if this happens, then do that, if that happens into that, but that now we have the ability to really list. And I was I did this life build in a YouTube channel recently. Here's a financial statement, analyze it, what's wrong with it. That's not something we had available before that. So definitely, that's something that we're more excited about. I think there are more and more tasks that a senior person typically was involved in require much more judgment, that now we can at least explore the opportunity to automate. And then I guess you can you can set your own parameters now to write, which is kind of like the the latest thing, right? Or it's not that new, but yeah, relatively new, where you can you can have a custom GPT and kind of have it pre populated with a lot of these prompts so that it can just kind of like digest whatever it is you do. And then you spoke to me at one point about like agents, I don't have a proper understanding of the difference between like an agent or like an AI agent and a custom GPT. Are you able to walk us through that? Yes. So actually, let me just walk you the way I see it as like four levels of AI. And I'll do it using just the tools we have with open AI as an example. So the first one is just chat GPT, the one that's meant for the user, right? The challenge with that is very useful. But it's a human has to interact with it to provide the data and to get the data and then process it. And it's also a generic model. It's meant to kind of do all sorts of things, all sorts of tasks for all sorts of people. So a level above that would be a custom GPT, which you brought up, where you can now have a specific prompt, a specific instruction that you're given it. So now you're making it a little bit more specialized. But still, you have to interact with it as a human to get the answers you want. A step above that would be within open AI is called a GPT assistant. And that's where we start to get into kind of like agent territory, I see really agent as a spectrum. But GPT assistance, it's like a custom GPT. But now you can use it in automations, and also work with some of the variables that are behind the scenes like temperature and other things that will affect the performance of the of the system. And a step above that, that we haven't found a lot of need for, but I think it will become increasingly interesting. It's actually fine tuning the model with your own data. And closing that loop. So that as you let me put this with a concrete example, I think I'm particularly excited about automating, coding, categorizing for exact transactions in kind of bookkeeping world. But with a human in the loop, what that means is that we have a model. And every time a human categorizes a transaction, the model learns from that, and tries to categorize the next one. So in that it can in that at that level, you have a layer where you're really almost like doing what chat GPT or open AI did for the entire world. But a much more scale, you just work in it, training the model with the data you have available. And that's kind of where you're getting into agent territory, where it's effectively like another employee, almost, that has the data that it needs to kind of can capture things and can kind of like also act on your behalf. Interesting. I like that. I like that breakdown. You are, by the way, I think you're just an excellent communicator of very technical topics to a less technical audience. So thank you for that. Yeah, I appreciate it. Yeah. And I mean, I think that that takes years and years of dealing with us accountants trying to hammer that in and it shows so that's great. That's the clearest I've ever had that described to me, which is cool. It's also interesting to understand that like, I only know of like levels one and two. And like, even at that, like, I'm a relatively tech savvy person, kind of, but like, it's cool to know, like, I've heard of agents, but I didn't really understand them. And I've definitely never experimented with them. But it's cool to know that there is actually truly so much more out there than like, we would necessarily know. Yeah, yeah. And ultimately, I think that that's the future where you have an agent that you over time give it more, quote, unquote, access to your data, and also more privileges to do things on your behalf. So I think that's, that's really the analogies is you're trying to build another human like entity, human like employee, but it just happens to be AI. And over time, maybe if it's if you're training it to be an assistant, over time, you give it control over your inbox and get it to at least draft your emails in your voice or, you know, categorize your transactions or whatever it might be. Yeah, I find that super fascinating. And the in your voice is is a good qualifier there. I find that to be like, particularly interesting where, and the more you write, the more content you produce, the more kind of ammo you have yet to be able to kind of like, get this thing to understand, like the way that you communicate. And I find that very, very interesting. And as a content creator myself, I'm like, wow, that's, that's actually really cool where the like, the more stuff you're putting out, the more that it can kind of digest who you are and be able to represent you, which is a little bit scary, but mostly exciting. Yeah, yeah. And that's why I also believe in like, trying to close the loop and have the human involved at some point. Because for the example of like, getting an AI agent to manage your inbox, I would love for it to do the draft. But then if I change the draft, I wanted to learn from that, and see that I changed it so that next time the draft has a higher chance of being the way I like it. And I think that will become more and more important. A lot of people are fear mongering around AI, I think that the human species has been incredibly ingenious in finding other work to do. So I'm not particularly worried about accountants losing their job or anything like that anytime soon. Yeah, an exercise I really enjoyed that someone talked to me about was like, try to find like old articles about like, internet, or, you know, just things like that, like other big leaps in technology and people like, oh my god, it's all over and, and, you know, what are we going to do and, and like, there's so many places I mean, up until maybe two years ago, but like, if we look at two years ago, so many places were like, desperate for talent. Right? And it was like, oh, but like, you know, using these code repositories on the internet and whatever, like, we thought that was going to like eliminate devs and it's, it really hasn't. So I think that we can gain a lot of comfort from history of, like you said, people will find something to do, like, you know, a lot of places right now are short staffed, and, you know, it might be in different avenues, but like, and, and there's also like a certain level of human touch that's never gonna, that's never going to be replaced too. So yeah. Yeah, definitely. Okay, so I have two kind of more questions I wanted to cover. One sort of puts you on the spot a little bit, but the other one, let's start with the other one first. So for the first one is kind of one of those things, you really get out what you put in with these kind of GPTs and AI in general. Where do you think people should focus if they want to improve their kind of prompt engineering? It's a lot of experimentation. I don't think there's a good answer, other than trial and error to an extent, right? Just test, see how it works and try again. I've found some resources that are good, even OpenAI has some resources on prompt engineering. But here's what I find interesting. The AI is trained to understand a human. The clearer your instructions are for human, the clearer the instructions are for the AI. So just focus on being, communicating clearly, and being able to specify what you want potential with examples in as much detail as possible, and you're well on your way. I think that prompt engineering is almost like a buzzword, but fundamentally is the ability to write a clear instructions for, in this case for AI, but it is effectively under the same principles of writing clear structures for another human. Okay, interesting. Thanks for that. And then yeah, the other thing I wanted to ask, oh, sorry, go ahead. I think that wasn't the answer you were expecting. No, not really. And there's so many tips and tricks, but it's cool to actually hear the real deal from you because some things people publish online, it's like, you know, oh, five tricks to blah, blah, blah. And it's like, yeah, their job is to generate clicks, right? So giving the explanation you gave, like people aren't going to click into that article, but it's probably closer to the truth, right? So yeah, definitely. Yeah. Okay. So the other question where I'm kind of putting you on the spot here is knowing what you know about big four transparency. So right, like we're a database, we collect data from people anonymously, and then I'm trying to kind of help the community understand like what you should be compensated and what your career path is, right? And so there's a flow of like an input form and then this kind of database and then the data is kind of presented back. In that, is there a place where you think I should be incorporating AI for either like a really cool feature or just like a general process improvement? Well, I don't know if the thing that comes to my mind is potentially filling gaps in the data, if any, if there are any. Yeah. Like imputing kind of missing data points for for someone, essentially. Yeah, like trying to infer. I don't know, have you changed your form over the years, for example, maybe there's a thing you added to the form that wasn't there in the past? Yeah, yeah, actually, some of them. I've added like job satisfaction, hours worked. Those are kind of the main fields I've added in terms of like salary and bonus that's been pretty steady. But yeah. Yeah. The challenge is that you're selling that data and you want to be as accurate as possible. Yeah, so to accounting firms, I am but to the to the user as well. That's that's not monetized. I have like a free, right, like to the user come into the site. That's a complete free suite. So I guess I could use it in theory to like, impute like based off of all these parameters, this is where you should be, right? Yeah. Okay. What comes to mind, honestly, I don't know if this will be useful is maybe a chatbot experience that you could have for the user, where they can sort of like ask what they want out of the data. And potentially, this is just me trying to think creatively here. So I apologize for the rough ideas. I don't know if you have something like this, but having a picture sort of like potential like career paths and increases in pay and how that could look like. Even trying to effectively paint a picture for a student or a junior person for what their, you know, next 10 years are going to be. And again, I'm thinking here, and maybe even giving that, helping them plan their finances around that, right? If they had a financial advisor that they now can know, hey, these are my expected increases in pay as I become a senior in this role of the cuff. But this is just some of the ideas that that came to my mind. That's it. I like that. I like the kind of advisor route where it's like, okay, cool, like, don't worry too much if you're behind on your savings, because most likely, you know, in the next three years, it's going to change dramatically and like, your career earnings curve looks very different than some other career earning curves. And so like, take some pressure off, right, or something like that. Yeah, I think that's interesting. Yeah. Yeah. And I think that most, you know, when you're working in financial advisor or something of the sort where you're trying to calculate how much you're going to retire with, it's almost like a fixed salary, and doesn't take account some of those potential increases in pay. So I think that could be an interesting use case where, hey, this is not just assuming you're getting an increase in pay. It's just based on data of what people will get paid as they progress in your career. Interesting. I like I like it. I like it. And thank you very much, Isaac, for joining me. Before I let you go, I want to make sure I kind of include this, you have a sheet with like a summary breakdown of all the practice management softwares, or I mean, maybe not all of them, but a substantial number of practice management softwares, which I'm going to link in the show notes. I think it's important for people to remember that like process improvement goes, you know, it's a very, very kind of broad spectrum. And so like, even just sort of having to help with picking the right software for your situation, I think is like massively valuable. It doesn't all have to be AI agents and things like that. Like there's a lot of off the shelf things that are ready to go. And even just having someone to do that little like last mile integration between those softwares can be massively, massively valuable for your firm. So I'm going to make sure that I link that in the show notes. And yeah, thank you so much for all of your time, Isaac. Thank you. I appreciate it, Dominic.