Build Your AI Muscle

Rebecca (00:00): Okay, this was a really big week in the AI world and we in our Slack channel for ALFA we couldn’t help but talk with each other about how this is gonna change the profession and it’s interesting because it’s is it gonna change how how is it gonna change the profession in the next month or in the next six months or next tax season or next year or three to five years and with how everything is happening

Rebecca (00:25): just in a couple months or weeks things are changing. It’s almost impossible to think about like three to five years from now. so we figured we would take some time to kind of dive into that today and just think about like what services we would be offering, what our what our team would look like in an accounting firm. Yeah. What are your thoughts, Charlie?

Charlie (00:47): It really was a big week. I mean, Anthropic dropped a bunch of new features.

Charlie (00:52): They brought co-work and chat together, they added projects feature, they added documents feature. And it’s kind of going toward something what I’ve been thinking about a lot lately is like just this one app, this one mega app where you can go in and do just about everything. And I think that’s where we’re headed. Like right now we work in twenty different apps. I mean

Charlie (01:14): just all over the place, right? And now both Anthropic and OpenAI are just starting to cannibalize these various apps. and I think that

Rebecca (01:22): Give me examples of what apps you’re talking about.

Charlie (01:25): well I mean like this recent update, right, where you can basically work within sheets, documents, PDF inside of the app.

Charlie (01:34): And then you can export out, like if you need to give a deliverable, but you’re not like working inside of those apps. You’re working inside of Claude to do that work and then you can deliver the work and export it out in those same apps.

Rebecca (01:47): Kinda reminds me of like

Rebecca (01:48): when practice management software started adding like email inside of the practice management software.

Charlie (01:55): Yeah. And you know, Claude can already do that right. It can hook up to your email. It can draft emails, it can just put that draft right into your inbox. So you can just go into your inbox, you can see the draft. I mean you can even send it if you give it those permissions, but

Rebecca (02:09): Right.

Charlie (02:11): so yeah, I mean I think and then open AI, their app is very good. It’s very similar where you can just do a lot of this work. You can open up a browser within the app, you can just do a whole bunch within these apps.

Charlie (02:22): and they’re just gonna keep getting better and better.

Rebecca (02:25): So what does that mean for the next

Rebecca (02:29): year like go go we have a few months till tax season right like it’s we’re about to enter quarter four and we got a few months to to prep like what’s our tax season gonna look like this year maybe it it’s not gonna change dramatically this year and then what is it gonna look like in one year or three years or five years and I’m thinking from like a staff perspective, a tech stack perspective and the services that we offer.

Charlie (02:54): Yeah, I think this coming tax season may not be

Charlie (02:58): a super like big change. But I do think it people are gonna be using these apps such as Claude, and Chat GPT more and more to do some of the work that they’re doing now. Right? They’re gonna

Rebecca (03:08): Mm-hmm.

Charlie (03:09): start doing some of that work within the app, maybe still exporting out and sending it to the clients or whatever they need to do. Now in three years from now, five years from now, it’s very difficult to imagine what that’ll look like. I mean five years from now it just feels like eternity. So

Rebecca (03:26): Are the companies like Anthropic and OpenAI, are they s saying what they what their goal is for the like do we know like what the vision is for for three to five years from now or they’re not they haven’t really disclosed that?

Charlie (03:38): I don’t think they’ve explicitly said it, but they’ve said it by the features that they’ve started adding. I mean OpenAI just came out with Astra for Law, right? And so they’re targeting Yeah,

Rebecca (03:50): So tell me about that, like what what does that do?

Charlie (03:53): so it’s basically just a specific model.

Charlie (03:56): trained on specific data, in this case, you know, law, legal, and specifically targeted at lawyers, right? And so I don’t know if you’ve heard of this other application called Harvey, but it’s a very similar concept and its valuation right now is massive. I mean

Charlie (04:15): they just

Rebecca (04:15): Yeah.

Charlie (04:15): raise I can’t remember what it was, but billions and billions of dollars, tens of billions of dollars. and their valuation is you know, around I think fifteen billion or whatever it may be. But basically just a large language model trained on a bunch of case law and it’s aimed at lawyers for them to use. So it’s like a specific version of

Rebecca (04:34): So it’s for the

Rebecca (04:34): professional to use, not for the client.

Charlie (04:37): Right, no, it’s aimed at lawyers. so

Rebecca (04:39): Okay.

Charlie (04:40): professionals and this Astra for Law is basically like a subset of Chat GPT, but really specialized to handle legal case law and for lawyers to use. And so, you

Rebecca (04:53): Okay.

Charlie (04:54): know, they’re going that route. Claude just came out with with kind of a similar version but aimed at finance professionals. So you can connect

Rebecca (05:02): Yes, I saw that.

Charlie (05:03): a bunch of

Charlie (05:04): various different institutions and see the investment accounts of a client and that one is aimed at the financial professional to use to help with their client so you got openai targeting lawyers you got anthropic targeting finance professionals and you know claude already has their like small business offering that they have and so I think they’re just gonna

Rebecca (05:26): That’s right, yeah.

Charlie (05:27): they’re just gonna continue

Charlie (05:28): continue targeting these various subset of professionals with these really specific models that are trained. Yeah, and I

Rebecca (05:33): Mm-hmm. There’s been like marketing ones, right?

Charlie (05:38): mean it’s just gonna continue to go down that route. I mean eventually there’s gonna be chat GPT for accountants, chat GPT tax,

Charlie (05:45): It just kind of seems like that’s a natural progression. I mean OpenAI in my mind seems like they’re more gung-ho on like being the app, like one app, where it’s all you need. You come in this app and you can do basically everything you need. And the rumors

Rebecca (05:59): Mm-hmm.

Charlie (06:00): are they’re gonna they’re coming out with their kind of personal agent next week. I think they were

Rebecca (06:05): Okay.

Charlie (06:05): trying to drop one this week to kind of compete with instinct and meta’s muse and X’s

Rebecca (06:11): Yes.

Charlie (06:11): Grokbot.

Charlie (06:12): you know, OpenAI acquired OpenClaw, right? So like that’s

Rebecca (06:16): Mm-hmm.

Charlie (06:16): been underneath. And OpenClaw was really big back in January, right? And it’s been a bit stagnant since they had a bunch of issues and they worked through some security concerns. but it hasn’t like really took off, I think, like they thought it would. And I think it’s because chat GPT has been working on their own version of it and they’re gonna release

Rebecca (06:35): I see.

Charlie (06:36): it. And so, you know,

Charlie (06:37): I’ve been kind of talking about these personal agents a lot and people kind of get confused of like why would I use a personal agent? I have all this work to do. I don’t see the use case for a personal agent. But I think it just kind of goes back to the mindset, right? If you can start using AI in other areas of your life, including your personal life, I think it’ll bleed over into your professional life. Because you’ll start seeing

Rebecca (06:59): It’s like strengthening

Rebecca (07:00): that muscle, like you’re just yeah.

Charlie (07:01): Yeah, you’re exactly

Charlie (07:03): you’re basically just building this muscle, this mindset of, I use AI to compile my grocery list and figure out what aisle they’re on, even though you may already know this information. Or you you did it to track your workouts or you just kinda used it in these various ways, it just starts kind of open up your mind like, okay, well I could probably use that in this side of my work or this type of engagement that I do.

Charlie (07:28): and so I think I mean that’s what I recommend because a lot of people are scared to use AI on client work because they don’t fully understand how it works, so they’re a bit hesitant to do it, to maybe mess something up or expose data or whatever may do. So I think it’s smart to use yourself as a guinea pig and still keep it low stakes, like

Rebecca (07:46): Yeah.

Charlie (07:47): creating these just kind of small little tools or dashboards or apps that you can have these personal agents use.

Charlie (07:54): use and or have them go and complete these like small minuscule tasks. It’ll just kinda

Rebecca (07:58): Yeah.

Charlie (07:58): again build that muscle, that AI muscle to where you can start saying, okay, now I’m feeling more confident using it right. I got it to do these basic

Rebecca (08:06): Right.

Charlie (08:06): things. Let me try and actually do it on this workflow or or kind of use it in this part of this engagement. And it’ll just kinda again build that confidence.

Rebecca (08:15): Yeah.

Rebecca (08:15): Well and that puts us in the shoes of our clients. Like if we’re gonna use it in our personal life, then we can think about like what what will our what are our clients’ needs and how for the finance side of those things, you know, how does how does our services tie in? For example, this morning I finally downloaded Muse and you know I asked it like straight up like here’s everything here’s a summary about me, what can I use you for? Like given my

Rebecca (08:41): life in my situation, like what’s the point of having you? Like, show me your worth. What can you do for me? And one thing it said was like, well, you have all these different businesses and bank accounts and stuff. So if you’re comfortable giving me read only access to these accounts, I can help you look at account balances and make sure things are going well. So because I’m comfortable experimenting on myself, maybe more so than a client, like you said, I did give it read only access.

Rebecca (09:08): to some of my accounts and it’s already gonna schedule something every Monday. It’s going to tell me what one of my bank balances is to help me remember if money needs to be transferred there or not. And that’s just something that has always kind of been lingering in the back of my head like, I really should be checking that. And it’s like someone’s gonna just do that for me now.

Charlie (09:28): Yeah.

Rebecca (09:28): so if if our clients could benefit from that maybe we do a little AI advising when it comes to the financial side or the tax liability

Rebecca (09:37): side of our clients’ lives and kind of support them with helping them figure out what tools like that they should be using. Maybe that’s one of the future services we offer.

Charlie (09:48): Yeah, I think

Charlie (09:48): that’s a great point because I’ve actually done this. you know, I think everyone now is probably getting emailed basically AI kind of assisted or generated emails from clients, right? And

Rebecca (09:59): yeah.

Charlie (09:59): a lot of times from the consensus that I’ve gotten, people don’t like that. Right. A because usually you get this wall of text, you know, yeah, they’re

Rebecca (10:06): They’re long.

Charlie (10:07): long, they start out with high.

Charlie (10:09): You know, they have the tail signs that is AI generated, you know, dashes throughout whatever it may be.

Rebecca (10:16): Yep.

Charlie (10:17): and you know, and a lot of times also the email is

Charlie (10:21): just NA, you know, there’s just junk in there, there’s non-relevant stuff, right? And so a lot of people are taking the stance of I don’t want my clients using AI. Well I’m I’ve taken the opposite stance. I do want my clients using AI. But just like I took a lot of time to get my harness set up right to where I can use AI really well and get really good output. I’ve now taken the time to set up my client’s harness. So like I’ve

Charlie (10:47): kinda explain to them, hey, you know, you sent you asked me this question,

Charlie (10:51): It’s not like relevant or like this isn’t a good idea to do, and here’s why. And the reason that your AI suggested this was because it doesn’t know X, Y, and Z that I know, So instead

Charlie (11:03): of me like gatekeeping this information, I want to give it to the client and say, hey, here’s

Rebecca (11:08): Genius.

Charlie (11:09): high-level, you know, nothing crazy, right? But here’s our previous conversation summaries, here’s detail about

Charlie (11:16): you this relevant your business whatever and here’s

Rebecca (11:19): Mm-hmm.

Charlie (11:20): some positions that we’ve taken here’s our goal right as we’re working together give this to your agent if you’re comfortable with it and most of my clients are they’re using AI heavily so like I’m trying to now take the time to get them set up in a better spot with their right harness so that they can now ask some of the questions of their own AI and get better and quicker answers because they have the context because they’re

Charlie (11:44): AI now has that context.

Rebecca (11:44): And I and I don’t think they’re

Rebecca (11:46): gonna be like, Great, now I can replace Charlie. They’re gonna be like, Wow, it’s really nice to have a human who I can ask Hey, am I doing this right? Is this safe? Is this what I you know,

Rebecca (11:57): you’re now in an another advisory, service that you’re offering is like AI advisory from a financial tax business, perspective. so, some of us might be listening and thinking, well, my clients do are not using AI, like my clients are not of that generation, you know, but three to five years from now, perhaps they will be. if so much has happened in a matter of months, maybe

Rebecca (12:19): we really do need to learn AI ourselves both personally and professionally and then not be annoyed when our clients are using it but use that as almost a sales pitch of like

Rebecca (12:31): Great, I’m so happy you’re using this. This can help your life quite a bit. It certainly helped mine. Let’s just book a, you know, AI checkup or whatever. You know, that’s actually a service that we offer now where I can get your harness and even explain what a harness is to you, get it all set up

Rebecca (12:49): it’s a waste of your time to send me a question that doesn’t make sense and go back at you don’t let want that either. So let’s take that time together. this might not sound realistic right now, but I could imagine a year from now, this would probably be a really smart thing to adopt. And that kind of brings me to what I think staff should be doing right now.

Rebecca (13:09): Because some of our ALFA members really are automating a lot of the tax work and bookkeeping work, to the extent that

Rebecca (13:17): It’s all happening. Like they wake up in the morning and things happened. things are happening a hundred percent without a human. Certain things, right? Like data entry type things or compiling document type things. so where does that leave the entry level staff that, you know, that were learning those things? And it brings me back to thinking about in school, like it was middle school maybe that we were learning about

Rebecca (13:40): how to use Google to research to write a research paper and know when it’s a credible source versus not a credible source because it was we’re not using, you know, like the card catalog in the library anymore. We have the internet now, but we have to know when we search something if it’s a credible source. I just remember them always saying credible source and we’d have to cite our sources and they would say that’s not a credible source or that is. So that was like a big thing that we had to learn. And so our staff, the entry level people, they’re gonna need

Rebecca (14:08): To learn with AI, is this a credible response that I’m getting? Just like that client who, you know, they didn’t ask the right question, clearly, if it was telling them, you know, the their AI didn’t have enough context because it was giving them information that didn’t make sense to you given their situation and what you know. so I think our team, we’re gonna have to teach them, maybe it’s even simulating with

Rebecca (14:32): examples that you have them test and then you look at what the conclusions they drew or how they reviewed the work and you help them see if you had asked it this question, it would have given you this. And you know, we kind of coach them through it and teach them how to use that second brain to really help them review the data entry work.

Rebecca (14:53): Because that’s what our clients are paying us for. they’re not like, Charlie is the best categorizer ever. Just the way he categorizes the transactions is better than anyone else I’ve ever found. It’s that, if AI were to categorize those, Charlie makes really good judgment calls about whether

Rebecca (15:08): things are put in the right spot and he can then, extrapolate that into helping me make business decisions and make me feel good about, you know, what’s going on with my business or, things like that.

Charlie (15:19): Yeah, in my mind I see a future where there’s this shared knowledge space, right? This shared context space. it’s probably gonna end up being something like an agent native portal.

Charlie (15:29): Right. You know, right

Rebecca (15:30): Mm-hmm.

Charlie (15:30): now we all have client portals, I upload something to the portal, client logs in, they get it, they pull it down, they upload things, I go in, I pull it down, right? Well, I think

Rebecca (15:40): Yep.

Charlie (15:41): that’s probably gonna end up growing to be this kind of environment where certain information lives, And my agent can both send and receive information from there, and so can the client. But what lives in there is

Charlie (15:56): I think gonna go beyond documents, It’s gonna go back to this context that AI create, The accountant creates, right, and then sends

Rebecca (16:06): Mm-hmm.

Charlie (16:07): to the portal. And what stays back is gonna be my work product, Like all my work papers, everything that I do. But I’m basically gonna have to use AI to pick out.

Charlie (16:19): what it thinks or what I think the client needs and send it to the portal. And so now what

Rebecca (16:24): That’s so interesting.

Charlie (16:25): lives in this portal is like call summaries, decisions that we’ve landed on. the clients our north star for this client. what’s this client’s goal? This is what we’re working towards for this client. So it’s gonna start it’s kind of gonna go beyond accounting and taxes in my mind. It’s gonna get get a lot more into that advisory space where we’re gonna say, okay,

Rebecca (16:44): Yeah.

Charlie (16:44): let’s think big picture

Charlie (16:45): Of what’s the goal for this client, right? And

Rebecca (16:47): Mm-hmm.

Charlie (16:48): now let’s work collectively to get there, I you know, we’ll do probably more the lion’s share of the actual work. The client is gonna have a if they’re a business owner, they’re gonna be managing their business. But

Rebecca (17:00): Right.

Charlie (17:00): you know, a lot of business owners still have these questions and they still want to think through some of these scenarios themselves, and they don’t want

Rebecca (17:07): Mm-hmm.

Charlie (17:07): me some of them send us every question that they can think of, others are more subdued, right? So I think the

Charlie (17:12): key is again creating this environment to where the client can go if they have some just random question that you know if they sent it us, we’d probably be like, this is ridiculous question. but they can still ask and try to work through some of these scenarios because they now have this context that they no longer have to

Rebecca (17:30): Yeah.

Charlie (17:31): come and ask us directly for. And so

Rebecca (17:33): Right. And sometimes we’re like,

Rebecca (17:34): Man, I wish you had asked us because you just put

Rebecca (17:38): a rental property in an S Corp or something and like if you had just taken the moment to ask us. So it also could help there be less messes for us to clean up when we see them for tax season or something because they had the time to think about that. And I know you’re saying that it could go beyond you know just documents, but you know we’re we’re kind of going into tax projection season right now. And you know the the the biggest pain point is collecting you know the pay stubs and the documents that you’re gonna annually

Rebecca (18:06): To help figure out, if they’re paid in well or if all that good stuff. And it’s like just getting those documents. Like, I just wish there was a way that I could directly just pull what I need instead of having to depend on the client to get it for me. But I understand that they probably don’t want to give me access to their entire life like that. But if I could ask their agent, if my agent could ask their agent to go grab, everything we could possibly need to make a really good projection, and same with tax

Rebecca (18:34): to prepare their tax return, that could maybe finally end the document-chasing nightmare that we deal with every year.

Charlie (18:41): Yeah, I mean this is for sure going to happen. I in my mind, I mean even an agent out of the box

Charlie (18:46): It can do a a very good job of figuring out what documents are needed by looking at a tax return. you know, depending on what’s all included in that tax return, some tax returns have these statements where they have they list out the W-2s and they list out a lot more detail, right? If you just have

Rebecca (19:00): Mm-hmm.

Charlie (19:01): a a a 1040 where box one wages is 250k, it’s like is that one W two, is that three W twos? We don’t know, right?

Rebecca (19:08): Right.

Charlie (19:09): But you can still get a lot of detail from that. And then

Rebecca (19:11): Mm-hmm.

Charlie (19:12): just where the models are today.

Charlie (19:13): Day and if you prompt them of like, hey, I’m trying to figure out what documents I need, they can ask you follow-up questions, right? So I think the future is going to be the client’s agent will likely tell the client what documents it needs for it to upload. And it can probably

Rebecca (19:29): Now that

Rebecca (19:30): sorry, that just makes me wonder, like then this is gonna sound really negative, but

Rebecca (19:34): Why even have the accountant at all? Just just kind of exploring this for a moment. But if the agent knows what documents they’re gonna need for their tax return, and the agent, hypothetically can prepare and even review the tax return, then why are they sending us the documents and having us do something that potentially can be fully done directly with them?

Charlie (19:54): I think probably it’s gonna come down to the person. Like how much responsibility that are they willing to give their agent? And I do

Rebecca (20:01): Mm-hmm.

Charlie (20:02): think that’s largely gonna be determined by age, right? The twenty-five-year-old today who’s gonna be thirty in five years, is probably gonna have spent those five years a lot in AI.

Charlie (20:14): And therefore gonna

Rebecca (20:14): Right.

Charlie (20:15): be willing to probably offload more to AI. The sixty year old who’s gonna be sixty-five in five years is probably not gonna be as willing to offload because they’ve spent sixty years of their life without AI. And so They’re gonna be different.

Rebecca (20:28): Yeah.

Charlie (20:29): But I also think that

Charlie (20:32): lot of people aren’t probably gonna get this deep with AI, a lot of clients, right now it’s about the personal agent, For the consumer, the regular person, they’re having them book flights, go and reduce bills, find alternative

Rebecca (20:45): Right.

Charlie (20:46): insurance providers, find refunds, buy things off Amazon. I think it’s gonna start with the personal agent.

Rebecca (20:53): Mm-hmm.

Charlie (20:53): And for

Charlie (20:53): a lot of people it’ll probably stop there. And so like will they want their personal agent to do their tax return? I mean, a small minority sure will. The vast majority likely might not.

Charlie (21:04): I do think they’re gonna be people that are gonna have basically fleets of agents. Like they’re gonna have a personal agent, they’re gonna have a tax agent, they’re gonna have a finance agent, and they’re gonna offload a lot more responsibility and autonomy to those agents. I think the key again is positioning oneself to where you can work really, really well with someone that is AI native and is is willing to prov offload a lot of this work to their agent.

Charlie (21:33): Because again, I think it’s gonna vary as to okay, the agent can do a lot. Maybe it can get them to ninety percent. Is that person comfortable letting their agent go a hundred percent of the way? I think that’s

Rebecca (21:43): Right.

Charlie (21:44): gonna be just a a much smaller population of people. And I think mu Yeah.

Rebecca (21:48): At least for the next few years, probably or

Rebecca (21:50): year. Okay.

Charlie (21:51): Yeah. So I think it’s

Charlie (21:52): gonna be more about okay, my agent is helping a lot in this, but I still want your involvement and I still want you to take me on the finish line and I still want your advisory. I still want you to be a part of the mix.

Rebecca (22:03): Right.

Charlie (22:04): and so I think, you know, we’re gonna have to really figure out as a professional who we want to target and then figure out

Rebecca (22:09): Right.

Charlie (22:10): what that looks like for for each one of us and then position our firm, our practice accordingly.

Rebecca (22:16): And maybe have like transparent conversations with our clients about AI. we know you’re using it, that’s okay. Let us be on the journey with you. Let us help give you some tips.

Rebecca (22:25): I don’t know, I think that would make me be cool, my accountant is like knows about AI and like knows is okay with me doing all this and wants to help me with it. I think that might be a good approach. so okay, we’ve talked about a lot. What do you think an accounting firm owner could do

Rebecca (22:43): this or even just an accountant, you know, anybody in the accounting world, what could they do this week, like this coming week, to just help move them along on this journey? what’s the next step? What’s their homework until you know they listen to our next podcast? what should they do to to try to get there?

Charlie (22:59): I mean, I think they really should be spending a lot more time in these various AI providers. I do think the best thing someone could do is force themselves to literally do everything through Claude, ChatGPT, Gemini, Copilot, whatever it is. Like pick a day and say, Okay

Charlie (23:19): I’m gonna use AI to do everything. I mean within reason, right? don’t have it do something you’re not comfortable with it doing. But if you need to draft an email, have the AI draft it. If you need to look something up, don’t use Google, don’t use the web, use AI. If you have to

Charlie (23:34): you know, try to purchase something. Use AI to figure out what that looks like. I mean if you’re trying to figure out a certain workflow, you know, use AI

Rebecca (23:42): Mm-hmm.

Charlie (23:42): to work through it. And then again, I I do think these personal agents are a great playground.

Rebecca (23:49): Yeah.

Charlie (23:50): In a sense that you can get it to do these low stakes items and it will just kind of open it up your mind to like, okay, wow, like that was actually a really good experience. It did just

Rebecca (24:01): Yeah.

Charlie (24:01): save me my insurance bill. Or it did just go and easily buy this item if you give it those capabilities. And it also teaches you about these guardrails. I mean you have to give it access to certain things. You have to give it access

Rebecca (24:12): Mm-hmm. Right.

Charlie (24:15): to your email. What does that look like?

Charlie (24:16): Like what all can it do? I wouldn’t do any of it blindly. I would do

Charlie (24:20): exactly what you did and be like, okay, here’s part my life or things I’m dealing with or what’s top of mind for me, and I would start there and then ask it, okay, what would that look like? What would that entail? What would that mean? You have access to my email, what exactly does that mean? You know, ask it these questions and then

Charlie (24:37): another good thing people could do is start using the voice feature on these apps.

Rebecca (24:41): Yeah.

Charlie (24:42): I mean ChatGPT has a fantastic voice mode where you can

Rebecca (24:46): Yes.

Charlie (24:47): go into the app in the chat feature down below right beside the input box there’s a little button that looks like a little voice recorder.

Charlie (24:55): And just go on a walk. Or anytime you’re on a walk, anytime you’re in the car, anytime you’re sitting around, any question that comes to your mind, I would just voice it. I’d open up the app and I would just start having a conversation. And because the conversation that the agent will have with you is thorough and well done. it’s actually a really good experience. And so

Rebecca (25:13): Yeah. Yeah.

Charlie (25:15): I mean I’ve started doing this, like I go on these long runs.

Charlie (25:18): Yeah.

Charlie (25:19): And I will talk to my AI the whole time. I’m just thinking through different things that are on my mind. And it’s just like having a conversation with, a thought partner about things that you’re

Rebecca (25:30): Right.

Charlie (25:30): dealing with. And then after the run, I come back and I I copy that entire transcript that we had and I’ll start working through some of those things we talked about. I mean, some of it is just kind of I throw it out, it’s not really relevant to anything, it was a one-off thought. But some of those

Rebecca (25:45): Mm-hmm.

Charlie (25:46): are

Charlie (25:46): like okay it was something that I’m dealing with that I want to work through now so then I’ll go into the actual either this desktop app or via the CLI and start expanding on that conversation that I had because those are kind of high

Rebecca (25:57): Not so cool.

Charlie (25:58): level thoughts and conversation that I’m having in the chat on my run and then I start getting in deeper once I get back to my desk and on my computer and start planning things out.

Rebecca (26:07): And I love that because I

Rebecca (26:09): think some people will say like, I asked it this question and it it gave me a very wrong answer, clearly. But if you were having a back and forth conversation and it was talking back to you

Rebecca (26:19): you could say well, that doesn’t seem reasonable. And then in like through a through a dialogue, a two-way conversation instead of typing, I could see being able to shape it to get it to answer things the correct way, would be lot quicker because you could explain yourself more and explain why it’s wrong and it could understand that quicker. And I also think that the two-way voice feature is such a great entry point for older populations.

Rebecca (26:44): because they’re used to like talking on the phone and and having a two-way

Charlie (26:48): Yeah.

Rebecca (26:48): conversation and it doesn’t require a lot of technical knowledge. It’s just download the app, press the button, and it’s gonna talk back and forth. And I’m even thinking of like my a hundred and two year old grandma who’s lonely. Like she then she could have someone to talk to, it could be really cool to use it, you know, professionally and personally as just a way to, have such

Rebecca (27:09): a natural, simple entry point into using AI, I think.

Charlie (27:14): Yeah, I think,

Charlie (27:15): I I’ve heard from a lot of people that they don’t know what to use it for. And it in

Rebecca (27:18): Yes.

Charlie (27:19): my mind it’s just crazy to think about because I have so much that I wish it could do for me and and that it does do for me. but I do

Rebecca (27:27): Mm-hmm.

Charlie (27:28): think the key is talking to it. And I think

Rebecca (27:30): Yeah.

Charlie (27:31): it’s easier to talk to it via voice than it is when you’re typing.

Rebecca (27:35): Yes.

Charlie (27:35): it’s less friction, it’s a bit more natural and where these voices

Charlie (27:40): models are at today they’ve just improved so much

Rebecca (27:43): Yeah.

Charlie (27:44): it’s way more natural and you can just push back I mean

Rebecca (27:46): Yeah.

Charlie (27:47): I think a lot of people their ego kind of gets in the way a little bit when they’re working with these models and they you

Rebecca (27:51): Mm-hmm.

Charlie (27:52): know the model will say something and it’s you know they say that’s that’s a dumb response well I mean just push back on it. I mean don’t don’t be afraid

Rebecca (27:56): Yeah. Right.

Charlie (27:58): to say hey I don’t agree with that and here’s why

Rebecca (28:01): Yeah,

Rebecca (28:01): like where did you get that number? Like, you know, someone was telling me yesterday that it told them like a proposal for a partnership return should be seventy-five thousand dollars. And you know, it’s like you close your computer and say, Okay, well I’m not using AI then because AI is wrong, you know. It’s like, well, push back. Ask where did you get that number? What are you basing that off of? maybe I need to give you a little more information. that doesn’t sound reasonable, let’s talk about it more. and I think some of us are having some

Rebecca (28:26): fatigue with the massive paragraphs and even if you say TLDR it’s still it’s a lot of reading on your screen. But I found that when I had the two-way voice conversation it was much more concise the nice lady talking back to me it was just not those long walls of text anymore which was nice yeah

Charlie (28:43): Yeah. I found

Charlie (28:44): that I found that as well. again much more natural, right?

Rebecca (28:47): Yeah. Yeah.

Charlie (28:48): and I think it’s just more palatable and more digestible when you have these voice conversations. And so that’s where I would start. I mean, I would literally

Rebecca (28:54): Yeah. Okay.

Charlie (28:57): force myself to use it for just dumb things. You know, don’t waste your time. In my mind it’s not a waste because

Rebecca (29:04): It’s experimenting,

Charlie (29:05): Yeah, ’cause I’m

Rebecca (29:05): you’re learning.

Charlie (29:06): I’m always trying to learn whenever I’m using AI and I think that’s how people should approach it as well, is I’m

Rebecca (29:10): Mm-hmm.

Charlie (29:10): not just using it to using it. I actually want to actually know why it’s telling me seventy five thousand. Like if I got that response, I would be super intrigued by that. I would be so intrigued and be like, that is crazy and I really want to know why and how you got this number. So let’s dig into that. that’s all I wanna

Rebecca (29:26): Yeah.

Charlie (29:26): know now is where’d that number come from? And then again,

Rebecca (29:29): Right.

Charlie (29:30): just refining it.

Charlie (29:31): Every time. Every time it does something that you don’t like, you don’t agree with, you don’t like the output, you don’t like the way it’s speaking to you, just tell it and say, I don’t like this because of this. adjust your context, adjust your memory, adjust the instructions, whatever you need to do to change it so that the output is more like this next time or it’s different. And so I again I think

Rebecca (29:51): I said that the other day to it,

Rebecca (29:52): like, stop bringing that up, let that go. And it said, okay, I just updated my memory and context for that. And it was like, it was that simple. I just had to tell it to stop doing that. And it knew what it needed to do on the back end to make sure I don’t hear anything more about that. So okay.

Charlie (30:09): And then the other thing

Charlie (30:10): I’ll say, the last thing I’ll say on this front is that

Rebecca (30:13): Sure.

Charlie (30:13): if you’re a firm owner, and you are using it in these various ways, I would really be going back to my employees and kind of showing. Like if you use it in some way and you build this neat little dashboard that, you know, it’s about your personal life, maybe. Maybe it’s tracking something that’s completely unrelated. But just showing

Rebecca (30:29): Sure.

Charlie (30:29): it, like letting people see how you’re using it, So that your f your employees can start using it again.

Rebecca (30:35): Yeah.

Charlie (30:35): Again, maybe on a lower stakes item in their personal life on various things.

Rebecca (30:38): Right.

Charlie (30:39): I just think personally that it’ll bleed over into the work. Once you start using

Rebecca (30:42): Yes. Yeah.

Charlie (30:43): it more, you start having these like mini little light bulb moments of like, that’s very cool. It makes you want to start using it more and start thinking about

Rebecca (30:50): That makes sense.

Charlie (30:51): how you could use it and implement it in these other areas of your life outside of just, you know, using it personal, if that’s where you start.

Rebecca (30:58): So model it for our team.

Rebecca (31:00): and share with them what what we’re doing so that we’re, teaching by doing and actually sh walking in the steps that we want them to walk. and just start with simple things, like even personal life type things, and ’cause all of that is gonna strengthen the AI muscle so that it starts bleeding into, I wonder if I could solve this annoying payroll thing that I have to do every month and cut the time down, for our firm, you know, it’ll

Rebecca (31:24): kinda lead there. So save time, save money, get to go home earlier, all good stuff.

Charlie (31:30): Yeah, I mean I do think the possibility is for

Charlie (31:32): almost endless at this point. I mean if you have you know the drive the desire and the willingness to try to figure something out and you know spend time working through it I do think that these models can get you pretty pretty close depending on obviously what it is that you’re working on. but yeah

Rebecca (31:48): Mm-hmm.

Charlie (31:48): I mean that’s where I would start. you know as one of our members in ALFA Brooks, he’s a partner. Yeah I think he has like 50 employees, you know, six partners and his other partners

Rebecca (31:56): Mm-hmm. Yeah.

Charlie (31:58): were a bit hesitant.

Charlie (31:59): And so he started doing things, personally using AI, and he would go back and show the other partners. And, they were very impressed with what he was doing. And they started to say, okay, wow. They started to see it. And then, you know, they started to adopt it as well. And their employees started to adopt it. And it just kind of took off at that point. You know, again, once people see it, it just kind of takes off.

Rebecca (32:19): Yeah. Yes. Once it’s not

Rebecca (32:22): abstract anymore, it’s something in front of you, then your head can start your mind can start wrapping around it a little more. yeah. Okay, well it sounds like we have our work cut out for us.

Charlie (32:32): just start using it. That’s all I can say. Just start using it. Start playing around with it. Start with low stakes items and then work your way up as you start gaining that confidence.

Rebecca (32:41): Yeah, love it. Awesome. Let’s let’s go do it.

Charlie (32:45): Alright,

Charlie (32:45): well hey, this is episode two. Episode three will be coming out next week. we’re kinda letting, the news and everything that goes down during the week dictate what we talk about in the next episode. But yeah, just start. That’s all I gotta say.

Rebecca (32:58): Love it. Awesome. Alright, see you next week.

Charlie (33:01): See ya.