540 First Impressions: GPT-5.1 For Accounting Firms

It's like, it's like Christmas morning every time you get a new model in your chatgpt just, just warm your little heart. You run down the stairs, you open up that little surprise gift and it's sometimes it's great, other times it's a little underwhelming. OpenAI just dropped GPT 5.1, the first kind of major step up on GPT 5, which was a very big update as we've covered at length. GPT5 unlocked a lot of really cool new use cases for accounting firms. But how are you feeling about this new 5.1 that just came out at the end of last week? I've got some good news, got some bad news, going to give you the rundown of what's new and exactly what you should do next. Let's get into it. This one kind of came out of nowhere. OpenAI hasn't been talking about new models and they just dropped this on, I think it was on a Wednesday. GPT 5.1 previously, the big kind of frontier, like very best model in ChatGPT was GPT5. I'll run you through what 5.1 does and then like I said, there's some, there's some good news and some bad news that we'll get to. So at a high level, what it does, two main differences. The first is it is supposed to be easier to talk to, a little more warm, a little more friendly. I know about you when I see this, I'm like, I'm kind of bummed. I'm like, give me like the good stuff. Like I want this for technical work myself. Not doing a lot of therapy with ChatGPT. I thought this example it gave was. They gave this example in the announcement landing page of the comparison of GPT 5.1 to the previous version GPT 5. They told ChatGPT I'm feeling stressed and could use some relaxation tips. GPT 5, the old model just says here's a few simple effective ways to help ease stress and goes into kind of a big list. GPT 5.1 says, I've got you, Ron. That's totally normal, especially with everything you got going on lately. What is your reaction to that? When I read that, I'm like, man, I don't know that I like want my technology to be that friendly. Thought that was kind of a weird example to include so it is easier and more warm to talk to. Second big thing is it is better at instruction following and they gave a couple examples of like very nuanced things like only ever Respond with six words. And if you've tried to get very specific, like in more useful ways than that, if you've tried to get very specific with ChatGPT before, you know that pretty quickly it just starts breaking all the rules. So better instruction following that is a very big deal on all of the benchmarks. Like it's not a huge step up, but we only got GPT5, I don't know, two, three months ago, so I'm not complaining. Like better instruction following right now that might be what we want most because it's very, very, very capable when it works. And then sometimes it just doesn't work in ways that surprise you. Still couple other odds and ends. It's said to talk simpler and use less jargon. They'll also let you give it tones of voice, so you can leave it as is the default. It has a new. It has a tone called friendly, efficient. So friendly tone called efficient. And they've now added professional, candid and quirky. Honestly, I don't really give a hoot because I use my AI to like get work done, not to play patty cakes with by the fireplace at night. Now, I said there was good news and there's bad news. We want first. That's a rhetorical question. Let's go with the good news first. The important thing here is that every time we get these like iterative improvements in AI models, the tools that we use that rely on AI get better. So for example, a really interesting thing right now is AI review, the notion that AI can function as like your technical review. We have a couple, couple companies doing versions of this for tax in the United States. Juno and Tax GPT, to be clear, like both of them very, very early on, that specific functionality will be like, really interesting to see that develop because that's something where you could see AI becoming very good at that in the near term. And normally that technical review function like, arguably that's like the most costly work for you to staff inside of an accounting firm. Does it ultimately replace an accountant? Probably not, but it probably saves somebody quite a bit of time. It definitely improves the quality of the self review that the preparer does. Right? That was a joke. The prepared isn't doing a self review, but with software that can do that self review for them, like that just means the quality of the work when it goes to the senior, probably vastly better. Right. And like engineers, like the people that build software have kind of been torn on this, like how much you hand over the reins to AI to kind of just figure it out. On its own versus how much you still keep it on rails. In fact, many things that we see that people even call AI features are, are not as AI as they would suggest. Because the reality is when you run a software company that's like helping tens of thousands of people and there's 30 people in the company, all it takes is that AI like doing something you don't expect 2% of the time and you've just got an absolute nightmare on your hands. It's why when like AI launch or when Intuit launches an AI assistant for QuickBooks, what you're expecting is like this cool open ended ChatGPT experience and what they actually ship is like something that really will just do like 20 things and that's all it's allowed to do. And when you ask it to do something, it just goes, okay, is this one of the 20 things that I'm allowed to do? No. Okay, so sorry, I can't help you with that. Right now this like cool, potentially open ended AI experience just becomes this sort of on Rails thing. Because right now the more you open it up, the more AI can kind of maybe operate, operate outside of what you intended it to do, which is fundamentally not how software is supposed to work and not even really how software companies are designed to operate. You want the tool to do the exact same thing every single time. Right. And so it's been like hard to develop with AI. It's why you don't have very many killer use cases still. Like AI meeting recorders, great, those have gotten vastly more accurate and we can kind of query our transcripts, pull specific information out of it. That's all really good stuff. But when it comes to more deterministic things, like financial things, not a whole lot happening yet, right? Like not much about how we work has changed. I'd argue the biggest stuff has been around like, like learning how to write a macro or scripting and stuff like that, using AI for that, where maybe we didn't know how to do that stuff before. But as more and more software is kind of open ended and just relying on AI models. And so you look at, you know, tax research tools, the review, the review tools that we talked about using AI to get, like review a tax return or a month end close in a more open ended way, every time you get these little incremental developments, those tools get better. And you know what don't get better? The tools that are not using AI. And so if you've got like legacy tools in your stack, if you've got very mature tools you're using. And honestly, like QuickBooks and Zero right now are very mature tools. And it's been a while since they've shipped something that I'm super excited about. And QuickBooks did make a whole bunch of noise at, at Connect their big annual conference with a bunch of new stuff that they want to ship and this kind of new accountant experience that will even cost extra money for the accounting firm. And you're talking about all these things that they're going to ship. None of it's like really real yet, unfortunately. And so these big mature platforms that aren't really using AI in any way, honestly, it kind of bums me out every time we get a new AI model. Or maybe, I don't know, maybe it just brings us one step closer to the future and somebody being able to like build something better, you know, that's why it's so important to be thinking about who you're hitching your wagon to. Like, am I using the platforms that are, that are doing meaningful things with AI right now? Because with every new model, they get better. Now the reality is behind the scenes, software companies use like a variety of different models. But in general, as models improve, they're either using a higher quality model or the models that were using before are getting less costly. So they can now like afford to use better models or use like more AI. Like have to be less lean in the way that they use AI because it's not as costly to do so. It's really interesting talking to the like kind of new crop of AI first companies and their vision for what they think is happening versus the traditional companies who are still saying we've been using AI for decades. Which honestly in most cases is just a red flag that they're like kind of refusing to acknowledge the exciting new stuff that's happening. Funny thing happened recently when I was at a conference with one of this episode's sponsors, Makers Hub. So Makers Hub, they're like an AI accounts payable platform. The party trick is they're using like AI vision models to do extraction at a level that was kind of never before possible with OCR technology. But I was doing something at their booth. Some of the customers came by. Like, it is one of those tools where the customers are kind of raving fans. But a funny thing kept happening. Like people kept throwing out these ridiculous document situations they have where they're like, I got this client, I have X, Y and Z problem. It's a total mess. We spend so much time on it and they just start riffing on it like with the Maker Sub team and they're like, oh yeah, no, I can see like you can do this potentially. Like our product doesn't really do that. But actually if we change this or that this could potentially work. The team is just like, they're just a bunch of like document processing geeks who, who are also AI nerds. And the product very much is like the marriage of those two things. And so what Maker Sub does best right now? Like the biggest, ugliest accounts payable stuff that you got, Man, I had some gnarly clients where Makers Hub would have saved us so much time. But if you got any ugly bill pay stuff, class tracking projects, inventory, manufacturing, stuff like that, definitely worth giving Makers Hub a look to look to learn more, check out the link down in the show notes. If your firm still runs on spreadsheets, sticky notes, or half a dozen different apps to manage work, there is a better way. Enter this episode's sponsor, Financial Sense. It is the all in one platform for accounting firms to manage everything. Workflows, client requests, capacity, billing and team collabs. All in one place. Automate recurring work, track deadlines and get visibility into what your team's working on. Built specifically for firm owners who want less chaos and more control. That's, that's actually an important detail if you're the type of firm owner that wants more chaos or if you want less control. This one, this one's actually not for you. Now, this December 12th to 13th, Financial Sense is hosting WorkflowCon, a free virtual conference built around the theme clear the chaos, find your calm. And me, I'll be joining for a live Q and A on the subject of smarter client comms, but comms of all kind, using AI alongside a bunch of other firm leaders. It's pretty like Rockstar. List of speakers they got. Register for Workflow Con via the link below. Start your free 14 day trial of financial cents using the link below with coupon code JASON10. Grab all that stuff down in the show notes. Okay, so the good news is every app that we use, every app in our stack that uses AI in one way or another probably just got better. Including by the way, ChatGPT, which is still like the ultimate Swiss army knife AI app for accounting firms. Here's the bad news though is man, the more people I talk with and the more I look at like the accounting industry at large and the people who listen to this podcast are like so far ahead. But we do struggle with a degree of this. The bad news is we're still, like, years behind on our journey of, like, effectively learning to use all the stuff that's already out there. Like, if AI development just stopped right now, we would be figuring out new stuff for years. People would be building this stuff into products for years if the models, like, did not even improve any further than where they're at right now. And so what honestly bums me out is that more people will talk about 5.1, this new model, then we'll actually run out and go test it. There's still so much more, like, fretting and talking and all of that than there is action. Because it is really hard to change your habits. And if you never built the habit of, like, working with an AI tool day to day, that's just a tough thing to, I don't know, get in, get, like, get into your head that it's always there and ready to help. And my best tip here is just leave chat GPT up on an extra monitor so that you remember that it's there. You'll find yourself coming to it more throughout the day and getting help from it. It's just kind of. It's wild to me how much conversation there is around AI that oftentimes just isn't followed with, like, action and taking the time with it. So best advice I can give you here just around, like, changing the habits of yourself and your teams to be leaning into this stuff and, like, exercising that AI muscle a little bit every day. Recognize that just breaking behavior patterns is hard. It's hard for you, it's hard for me. It's hard for all of us. But this is what makes it doubly important to find the people for whom it's easy. There's a lot of folks right now that have very naturally taken to AI in a way that most of us have not. I would also argue that most of the folks coming out of school right now are way more adept with AI because they just use ChatGPT to get through school. And so they are, like, to them, being productive is this totally different experience of, like, combined personal thought, like, with AI thought, they just have, like, a wildly different way of being productive than most of us do because they just went through high school and college with AI tools at their fingertips. And if there's, I mean, to an almost problematic degree, right, where it's like you take that tool away from them and do you potentially, like, not know how to be super productive without it even, like, the perfect world is between those folks and between the gray hairs who are having a really hard Time breaking those habits and making the time to use AI. And so if you have a team that's maybe not as far along as you want here, let's figure out how to augment your team with somebody who can bring that energy to the conversation. Somebody's got to kind of be the battery. And if you don't have that person internally, consider that in your hiring, like that's a great thing to, to look out for. If you still don't have it. See if there's somebody external that you can bring in that will join in just the conversation with your firm on a regular basis and create like a systematic way, a forcing function to ensure that your team is getting that contact high from that high energy AI person I would say every single week. So for most firms that looks like, like a standing weekly meeting where we come in, we talk about what we tried that week, what worked and what didn't work. This is going to be really easy for some people, it's going to be really hard for other people. And it's why having that sort of self appointed champion, the person that it comes easy to and is very enthusiastic about all that stuff, having that person in the meeting is going to be a lot of weeks where that person carries the meeting. But the value is everybody else is just getting exposure to that conversation and picking up a little bit week by week. So if you are, if you are one of those people who will sooner listen to a podcast about what's new in GPT 5.1, then go out and test it and play with it. Just make sure that you've got some sort of forcing function in the weekly schedule to just be doing a bit of that and ideally doing a bit of that with your team. There's going to be some weeks where you've got lots of time to do that stuff, other times where you don't. And on the weeks that you don't, that's the value of being able to take 30 minutes with your team so you can see what everybody else has been doing. And even if you weren't like elbows deep in doing AI stuff that week, you can still get inspired from the other folks who spent a little more time with it on that subject. Actually got a great question, a voicemail somebody left me. Let's do a bit of accountant support count and support operators are standing by. Okay, here's the voice Mel. So as a cpa, I think becoming fluent in related fields such as law and or finance may be an advantage to deliver clients insights. AI and competitors cannot Provide at the short term. Obviously, you know, no one wants to become a jack of all trades and master of none. So the question remains, how to position yourself to be relevant for the future. Did we just. What I'm about to say might be really insulting. Did we just have our first voicemail from AI about AI? Listen to the end of this again. So the question remains how to position yourself to be relevant for the future. I don't know, man. That might be AI, but now I'm actually not super sure and I feel kind of bad. Okay, still, good question. Do I go out and try to build a whole bunch of other adjacent skills? I'm going to be an attorney now and a CPA and a finance. All these different things. Is that going to ultimately protect me from AI? Here's my take is it feels like we're trying to outrun AI rather than, I don't know, outsmart it. Like, find the thing that AI is never going to do. Because if AI is so good of an accountant that it makes me go and become an attorney, can't you sit, say the same about AI being good at attorney stuff? So I don't really see a world where I can amass enough skills to stay ahead of AI, because AI will always be able to amass skills faster than me. What I get excited about is like, what is the more precise application of my expertise? And maybe that looks like, you know, combining several of those skill sets. But simply amassing more skills, I don't think helps you to outrun the robots now. Worth considering. When all else is equal, when everybody's using the same tools and everybody's using the same AI, the most capable, the most experienced, the most knowledgeable people will still use AI better. And take it further. If you look at accounting firms today, accounting firms all run on the same tech. Like, there's an ecosystem of, you know, 200 tools that make up like 95% of what U.S. firms use. And the same applies in any other region. If somebody had seen that before the age of the computer, and they're like, everybody's just going to be using, like the exact same things to do accounting. Well, then won't they all look identical? That's not the case at all. Right. All accounting firms are vastly different. You can look at two accounting firms that use the exact same tech stack and they're completely different. Is it going to be any different in the age of AI where we're all using the same tools, but we're wielding them differently? Worth remembering, like, at the end of the day when all the tools in everybody's hands are the same. The Delta is still the humans, it's still how they use them differently. Now AI will almost certainly introduce new ways that we support clients. Like right now, we are, we kind of collect just anything and everything that entrepreneurs want us to do that they don't want to do. Like, how many times does somebody come to you with some random weird one off regulatory thing or something that is, you know, the business's annual renewal or just these weird off the wall things that you've never seen before? Annual renewal, that's a bad example. Everybody's got that. But some like weird little nuance thing where they bring it to you and they're like, hey, hey, did you do this for me? And you're like, brother, I've never seen this before in my life. This is some sort of like beekeeping compliance thing in King County. Like, okay, okay, okay, well, can you get that knocked out for me? And you're like, I'm a bookkeeper, man. You know what I do? You know what I went to, I have a master's in tax. They don't care and we do it anyways. I have a hard time imagining a world where you get a whole bunch of politicians making rules and laws and all that stuff, and AI gets so good that just all the blockers to being an entrepreneur go away and they don't have any more messes they can throw at us. That seems unlikely, but there is a world where we communicate what we do poorly and we like set up camp doing work that is, is probably soon going to be eliminated. So if, if you communicate your value as like, we do a bunch of bookkeeping, we classify all your transactions, like, yeah, that stuff's probably going away pretty soon. If you communicate what you do more through the lens of what we were just talking about, where it's like, hey, we work with beekeepers and we're going to help you work through all this stuff that's standing between you and spending more time with your bees. There's almost certainly still going to be a lot of that stuff that gets in the way of entrepreneurs doing what they're actually trying to do. Those things in all likelihood will change in an age of AI, they're probably gonna change fast. That's kind of the one thing with AI is things have always been changing. It just feel like, feels like things are changing faster than ever. But if you send a client a proposal and the service items in that proposal are, we're gonna reconcile three bank accounts a month that have between 300 and 350 transactions, that's something that's going away, right? If we position ourselves and our value as something that transactional that. That does feel like you're living on borrowed time. Because the moment that they work out how to automate those 300 to 350 transactions per month, they don't have a need for you if you are instead helping them navigate how to pull the accounting system together, make sure all the data gets to the right place, and be reassessing what are the right systems. In a changing AI age, where the right tools are changing more frequently now and then, figuring out how do all those tools come together and roll up into the reporting that has to be done for the tax return and for this compliance thing and that compliance thing. If that's what we're doing for folks, then I have a hard time thinking that stuff's going to go away. We're coming off a week of live events. It's why I sound like I just drank a pitcher of gravel talking with someone in Oklahoma City. Tell me about oil and gas stuff. Evidently it's got like its own sort of type of accounting that nobody wants to touch. And you can go in and work with one of those companies and command 30 to 40 grand per month doing a very specific, very nuanced form of accounting. Talked with another guy. They. They finally decided to double down on Legal. So they're niching into Legal did this whole rebrand and he's like, man, we just finished our rebrand and now we see like six niches within the legal niche that we're pretty sure look juicier and we want to follow that. Now, should we do another rebrand? And when you are like that far into the weeds and that much of an expert within a specific domain, those are, those are the folks. And that's the type of expertise that I think is most protected from general AI solutions getting really good. It's my favorite thing about niching down to like the. You never know what the next level of depth is. And the reason why you go to the first level of depth is so you can find the second level of depth and so you can find the third one. With. With each level, you're solving more rare and more valuable pains, which means you can charge more for what you do. Oftentimes it means it becomes easier for people to find you because you're the only one that does that thing. And the only people who do that stuff are a little more connected, but in the hands of such a specific expert like you, you're gonna be able to do wild things with AI. And the future either looks like a bunch of humans using AI in ways that is unique to them in the same way that we use software now in ways that are unique to us, or AI just does it. And if we truly get to the point where AI just does it, I mean, everything changes. That's like, I mean, we're probably playing a lot more pickleball. I don't know that I'm actually afraid of that timeline because if it actually gets to that level, that might not be the worst thing. I got an eight year old daughter. When I first came out of school, I was working for a guy doing 60 hour weeks year round up to like 100, 110 hour weeks in tax season. There's a lot of people who will do that stuff like their whole professional lives. I'm not ready to like root against something that can make at least some of this stuff go away. Right. And somehow the human condition is like, even when that stuff goes away, we seem to backfill it with something new. So to bring it back to the question is, is the solution to build more and more skill sets? I mean, I think the solution is to continue to grow, but this sort of growth to me seems like growth of width learning, like more things, going a little bit deeper in a bunch of things. I don't know if that's the answer as much as going really deep into specific things is more the answer. It seems like AI. I mean, AI is already a tremendous generalist. It's smarter than any of your friends while not being as smart as any of your friends at what they actually do. And so I don't know that I want to out generalist a generalist AI. It's a good question though. You got any questions, drop me a voicemail. You'll like my voicemail message. Spoiler alert. 516-980-4968 man, it has been a wild week. Events in Austin, Oklahoma City and Phoenix. They were so, so good. Got a bunch more live events coming up as you're hearing this. We're doing a tour of the west coast after Thanksgiving. We're doing the Midwest. After that, we're doing the east coast. It's been a ton of fun. You've been hearing me plug it on the podcast a bunch lately. But come out to an event, you'll have a blast. Meet a bunch of really, really cool people and if you're doing anything new with GPT 5.1 man. Tell me about it. Drop a comment if you've really seen like a meaningful delta and what it will do for you versus the previous model. That's all we got for today, gang. Thanks for being here. I'll catch you in the next. A heads up for my podcast friends. We put together the biggest, scariest project my team has ever worked on. A big old run of live events around the country headed up until year end to ensure that your 2026 is just the best year ever. Let me run you through the cities real quick. We're coming to Austin, Oklahoma City, Phoenix, Seattle, Portland, San Francisco, Los Angeles, Atlanta, Nashville, Chicago, DC, Philadelphia, NY City, and Boston with some super cool half day workshops that roll into dinners. You get CPE for it. It is an event that is built for your entire team to get the whole squad feeling inspired so that if you are a firm leader, you're not feeling like you got to put everything on your back. Headed into 2026 gonna be a super fun experience. It's gonna be a surprise guest at every city. We're gonna record an installment of the podcast live at each of these events, you're gonna be part of the podcast itself. What? And man, I would love to see you there. To learn more, head over to jasononfirms.com events and I hope to see you there.