AI in Everything: Maximizing AI Within Your Existing Platforms and Tools Part Two

Ray Sidney Smith | 00:00

Are you ready to manage your work and personal world better to live a more fulfilling, productive life? Then you’ve come to the right place. Welcome to ProductivityCast, the weekly show about all things personal productivity. Here are your hosts, Ray Sidney-Smith and Augusto Pinaud with Francis Wade and Art Gelwicks.

Ray Sidney Smith | 00:17

Welcome back, everybody, to ProductivityCast, the weekly show about all things personal productivity. I’m Ray Sidney-Smith.

Francis Wade | 00:24

I’m Francis Wade.

Art Gelwix | 00:26

And I’m Art Gelwicks.

Ray Sidney Smith | 00:27

Welcome, gentlemen. Welcome to our listeners and onward with this episode from our last episode where we were talking about basically how AI is being embedded in all of the tools that we currently use. It is becoming… Somewhat of a commodity feature in many of the tools. And I think that there’s a value to that. And so we’re talking through what that really means and what So at this point in our discussion, I want us to turn our sights on the more practical ways in which we can use these embedded tools. So that we can actually figure out where in our productivity stacks the tools can be most helpful. And I always think about it from really the simplest of things, right? And journey of a thousand miles begins with the first step, right? And I think that most people want to change the world with AI. And I just want to send the next email. And so I just really want it to do very simple things to augment my life. As I frequently say now to audiences, technology is there to empower human endeavors. It should always and forever be a tool to empower human endeavors. And so I don’t need AI to live my life. I need it to augment my life in these little small ways that help my life be better and the lives of people around me. So let’s talk through this. What are those ways in which you’ve implemented AI inside of your embedded work? AI tools within maybe Gemini and Google Workspace or within another tool or Copilot within Microsoft 365 or even ChatGPT or Cloud, which connects to many of the Microsoft products?

Art Gelwix | 02:16

I’ll start with a couple of examples. The application of AI within our, what I want to say, our daily driver apps is I think has caused some consternation for a lot of people because It’s like this is a new feature, but I don’t know what to do with it. I don’t know how to derive a value. I’ve approached it from the standpoint of start with what you’re doing already, which is looking for information and synthesizing information. You need to use it to leverage the information that you already have that you know is sound, sane, and… Validated So for example, within Copilot, within Microsoft space, if you’re licensed for that, you can use it looking at your document assets. You can use it within things like Excel to generate new formula structures, things that you’re already familiar with. Within capacity, the tool I use all the time. Being able to generate a chat around a topic that creates the note base that I validate that extends on the thinking around a topic that I’ve been working on, which integrates then into scheduled planned activities that I need to pursue. Why is that so critical? One, because you’re doing something that you are comfortable to do. But verify what the AI system in that tool is giving you back. You’re just not taking it at its face value. You look at it and you go, yeah, that’s not right. Or yeah, that’s really good. The second part is to look at it as search 2.0, to be able to go past that initial search prompt to say, hey, find me this stuff. But then get into it further. What else about this thing? Tell me more about this thing. Let’s expand on that a little bit further. Getting comfortable with that process in the tools that we already use is a safe space to do it. Because again, we’re familiar with what the end result should be. If we start to get familiar with that in just generic open spaces… We’re not entirely sure what the result is, so we either have to take it at face value, Or we have to kind of doubt how things are coming back, have that constant question. I think back to the early days of, say, Google or Ask Jeeves, if you really want to go back into the Wayback Machine, is a great example of that. When they first started entering into the world of natural language questions, that was just earth-shaking. Everybody thought this was the coolest thing ever. I didn’t have to put in keywords and connector parameters and ands or ors. I could just We just have the ability to ask more questions. We have the ability to have personas and structures that look at things a specific way. But starting from an application that you already know, When you look at the result that comes back, you can go yeah, it did look at it the right way, or I need to tune it a little bit better, or I need to adjust accordingly.

Ray Sidney Smith | 05:20

I found myself going into Google Drive and now having these conversations with the files in Google Drive. And it’s actually quite helpful to have that embedded conversational component that is very different than doing an advanced operation search, right? And I know all of the advanced operators in Google Drive, at least the ones that are publicly known to folks. And I use them very effectively to find things. But it’s been really helpful to have a fuzzy search approach when I know that file is in there, I just don’t remember X or Y about it and be able to just natural language ask for it is helpful. And then the other side is when you want to be able to say, “Okay, everything in this folder I want to talk about everything in this folder in kind of the same way you would with Notebook LM, which is a little bit more grounded in the data. Now you are similarly grounded in the data that is already in your system. So you didn’t have to go to Notebook LM and add each individual file. They’re just there and you’re able to go ahead and engage with it. So it’s really powerful.

Francis Wade | 06:27

There’s a mindset shift there. I believe is coming, which is away from changes being driven by what’s available as opposed to what’s needed. I’m riffing off of what Art said. But he said, you got to start with what we are doing already. And I think we should almost be skeptical or stingy about a new possible change, let’s put it that way. Simply because partly because there’s more changes that are being suggested to us that are being advertised to us than we can possibly implement in a lifetime. So this is not going to get better, it’s going to get worse. Decide to play the odds and say, you know what, I can’t implement all of the embedded AI options that are available in my I don’t have enough lifetime to do it. And the question is, do you experiment until you find something or do you take? שעוד פרק, שזה I’m not taking any suggestions whatsoever, which is a bit. Draconian, but… I’m going to look for the one or two changes that are going to move the needle the most in my productivity. Which would mean that I need to have an understanding of the bottlenecks to my productivity and focus my efforts on them. Believing that if I “Focus on the few.” they’ll move the needle the most, that there’s some Pareto effect in here somewhere. And I somehow believe that our future is going to be Less raw experimentation and more diagnosis, analysis and targeted searching. And again, partly it’s because We’re going to have to say way more no’s than yes’s to suggestions that we include an embedded AI in our productivity stack. We’re going to… 99% of the time, we’re going to, we got to say no. We got to, but the question is. Where do you get the Pareto effect from the 1% and how do you find the 1%? How do you find a few things that are going to make you 10 I believe that there’s going to be a shift away from experimentation and towards Self-diagnosis. – Quantified self, but quantified self with like a diagnostic self. That is part of the quantified self. It.

Augusto Pinaud | 08:44

Is not a reduction on the experiments actually an increase on the experiments, what is going to be or what they’re going to be, it’s a lot more effective. Because when you think 10 years back. If you wanted to test to play to improve. You did not have the data. Or you have some of the data, your understanding of the data. Now, The artificial intelligence has all the data. It can tell you where you’re strong, it can tell you where you’re weak, it can tell you much faster that you get aware of that. And then what it’s going to allow you to do is to analyze the data and help you identify those possible things that can do the Pareto Principle. So instead of you testing, okay, let me see, I’m going to learn how to… Get my inbox to zero, You will say, nope, your e-books collection is not the problem. You collect plenty. Your problem is you need to organize that thing. And then now– You, instead of focusing on what you what gets your attention or what you think you can bring progress, the data or the analysis of the data is going to bring you to that place. So If you think on the five steps of GTD, okay, collect, organize, it’s going to allow you, hey, it is really more collection that I need, or it’s more organizing what I need, or it’s more doing what I need, okay? And be able to identify where that is. Same thing when you want to improve The skill. And when you want to improve a skill, it will not get you to the master level. But it will get you to see it will be able to apply what kind of work for a bunch of people, not only you, And it will give you that path to competence much faster than what you have been able to do before.

Art Gelwix | 10:53

It’s interesting as we talk through this, two things that pop to mind. One I almost think that the implementations of AI within the applications that we’re using on a daily basis are unintentionally moving us away from better organization of our information. Because we don’t have to. We can just say, Tell me this, do this thing, take this block of files And give me the knowledge I need from it. We don’t have to apply a structure. Letting the system arbitrarily or brute force it. Just an observational thought. Second, though, is I think there’s a very tight analogy to for where AI is going specifically within our own applications. To When we look at the evolution of cars and trucks over time. When you had the first vehicles, it was a vehicle. That’s what it was. It had four wheels. It had a seat. It kind of had a steering column. That was about it. And it did a thing. But as time progressed, we got into variants of it. We had specialized ones to carry multiple people, one to carry lots of things, one to… Bounce with hydraulics. All of that specialization. Still came back from that base understanding of this is a thing we need to do, but now we need to do it in a specific way. Cars actually started to turn into the equivalent of personas. Which is, Fascinating. But I think the more fascinating part is we, I look at this as we are really at the transition point of going from a manual transmission, a stick shift, to an automatic. We have been doing manual transmission work with our data for years. We know where the things are. We shift this and then hopefully the engine keeps going and we go faster. Now we’re getting automatic transmissions. Now we put it in drive. And we’re trusting that. That the vehicle is going to Put us at the right speeds at the right times in the right ways to get where we want to go. Sometimes it works, sometimes it doesn’t. Sometimes we have really good versions of it. And sometimes like CVTs, where we have really terrible versions of it. But at the end of the day, we’re still… We’re starting to separate ourselves again from in this case from our information Automatic, you’re separating yourself from a lot of the driving experience. Here, you’re doing the same thing from your data. You could literally set up a system. Within any tool, any good knowledge management tool that is pulling in a constant feed of information from the web, gathering all of this data because you have tasked it to do it, and you’re asking it to synthesize and summarize that and provide it to you as an encapsulation, but you’re never seeing the origins. You’re never seeing that authentic experience of what that information was. And whether that gives you a better experience or a worse experience is really going to be up to, in this case, the quote vehicle that’s getting you there. But As someone who writes, and we all do this, all four of us on this podcast, as a writer too, I struggle with that. Because the synthesized information that I put into my writing is not my writing. There’s a narrative and emotion and a context and a structure that’s there. But if you ask an AI to say, summarize, I worry we are getting to a Cliff Notes version of the experience of our information. And if we’re looking to process mass amounts of information, then yeah, the Cliff Notes version is fine. But if we’re truly looking to understand the information we’re gathering and work with it. I’m not sure that’s taking us to the same place.

Augusto Pinaud | 14:51

There is an interesting thing. As we are experienced as a promise, right? History, while living history is very difficult. Okay. If not impossible, but if you think about it, This is not different than… When the search engine went from completely incapable to somehow capable. There was a time in email, in Outlook, where if you do a search, you need to be almost precise. You need to almost search the exact set of words for that thing to find you an email. Otherwise, You couldn’t find it. That’s exactly what you’re seeing here. It is true. Okay, dad. You get the information, but.. The ability to search where that information has changed. There was a time where I collect ton of articles, okay, that I wanted to read it again, that I wanted to have, that I wanted to make sure that I kept, because finding again that article was going to be Impossible. Okay. With The searching capabilities now It’s faster for me to search again than to search into my own files. Okay? Those articles. It’s simply better. And… That, I think, is part of what we are seeing here. Again, it’s the same… Concept just applied in a different way. Do. This improvement. You are going to see how our search abilities are enhanced, not by us. Enhanced by the system, how the artificial intelligence can tend to be pretty accurate. So that removed upon a friction from that, okay, if you are looking right now to do a research, okay, it really The friction The amount of friction that you find It’s very.

Ray Sidney Smith | 16:55

Low. I’m going to be a little bit of a contrarian here. I agree with most of what everyone is saying here. I will note a couple of things that I think are just good for us all to keep in mind. One is that this kind of AI – basically LLM-empowered chatbots. Embedded within our tools are highly capable. There’s lots that it can do, but it’s rather stupid. And so you have to keep this duality in mind, right? Which is that it can do all these really fun and interesting things, and it can do it really poorly about 30%, maybe sometimes 40% of the time. And you’re baffled why it does that. And FYI, the reason is because it was modeled after humans. Literally, it’s modeled after human thinking. And so… First, kind of mild disagreement, maybe it’s not a mild disagreement, it’s just fundamentally I have a concern that most people really haven’t organized their source data in a way that helps the AI Help them. And so there’s this problem that if you For example, For it. Years, for as long as I’ve been using Evernote, I have been collecting and organizing in a very specific way so that all of my Reading notes for every book are organized into specific notes. They’re tagged, all of that stuff. Now, when I use Evernote’s AI Assistant to engage with those reading notes, they are structured in such a way that the AI assistant has no problem giving me back what I want. Now, if you take that same scenario where you’re trying to engage with those notes in a practical way, and you take someone who has just basically thrown haphazard, half-written, thoughts-in-the-margins type notes into their notebooks, and they’re all over the place. Now the AI has to do a lot more searching and it can’t quite find all of the context because it’s not in the same notebook. It’s not all tagged correctly. It’s going to give them back garbage. And so most people have not done that work. And I just, that’s not a casting an aspersion at others. It’s just most people don’t have the OCD nature I do to want everything to be organized the way I like it to be organized. That does not help them. And so AI is a tool that requires consistency. Focus because there’s an upper limit on what it can remember. People keep thinking that it has unlimited memory to find everything. You know what it does? It very confidently says at some point, nah, I’ve looked at enough because compute is capped, right? So it doesn’t have enough compute power. The captain’s office says, whatever, they’ll never know, right? It’s the overconfident intern, right? They didn’t look at the source documents. I did. And so they’ll take my answer as being the right answer. So I feel like we need to be very mindful of that fact that it can very confidently be fool us into thinking that It has organized. It has made sense of things that it actually hasn’t fully done its homework. It doesn’t get to 100% most of the time. It’s giving us this perception that it has, and that is very dangerous in certain circumstances. There’s.

Art Gelwix | 20:02

A reason why libraries were so well organized. And there’s a reason why librarians always seem so smart. Because the library was so well-organized. If you were to take a librarian… And take all the books off all the stacks and dump them into a big pile. What is your expectation of that librarian, Abe? Of them being able to assist you.

Augusto Pinaud | 20:24

You being the OCD, you’re going to get A lot more. Out of the systems. No question about it. But.. Remember, the person who, the non-OCD person, the non-very organized person, the non- detail-oriented with this capture, they still need to find the information in real life. And yes, it’s going to hit the cap. The same as this person is going to go. If you give them 20 notebooks and say, find a note in there, okay, let me look at that. I can’t find it. That’s what the human, it’s going to be. In the years where you needed to search into your own notes, that’s what people did. And reply, sorry, I missed, I lost the notes. It was like, no, you didn’t search. It’s the same answer, maybe a little bit more sophisticated. Okay. But, It is the same, but it will still… Amplify what these people have, it will still, there is something, have a better chance to find it and use it that it doesn’t. So The principles of computing continue applying. Garbage in, garbage out. What is the quality of the garbage that you are putting in there? If you’re OC, most likely it’s less, but if you are more disorganized, more than normal. It is the same.

Francis Wade | 21:48

I think there’s a separation that We are… Drawing between the things that AI is really good at and the higher level stuff that only we can do. So the example that came to mind was 20 years ago or how many years ago when Kennedy gave a speech Did Kennedy really give that speech? I don’t mean that, did he? Read the script. Where did the thoughts and the words come from? So at a certain level, you might say that He had 10 interns. He had 10 confident interns or speechwriters, right? And they did the speech. And what they did was they drafted back and forth and then they and sent a draft to him and he marked up and he said no and he said yes and then he decided that here’s the part I’m going to go with, and he delivered the speech. But so what AI is really good at, from drawing the analogy to us, is that AI is the 10 confident interns. Who produced the 10 confident drafts. And to pick up on what Art was saying, Today we’re struggling with If I write something and I use the input of 10 LLMs, or 10 drafts from AI. Did I really write it? And my answer to Art is… We’re uncomfortable quite saying that we wrote it. Right now, but the answer will ultimately be yes, we wrote it. Yeah, we had 10 interns and we’ve given stuff to interns to brainstorm and give us input on things that we write. But at the end of the day, We are the coordinating, we are the coordinator that takes all of these inputs. And even if the interns are expertly organized, and they have information and they have libraries and they have everything. It’s a different level of intelligence. Duh. Intelligently organize their inputs into a final output. So. Turns out, I think, that what we’re seeing is that The AI is really good at the 10 confident intern level. But so far, It’s no good at the delivery level, at the coordinating level. Recently, there was an article in Harvard Business Review in which they tried to get AI to develop corporate strategy. By basically throwing it information and asking it for a strategy as an output. And predictably… That AIs did a terrible job. They were awful. They couldn’t coordinate anything. It was for crap. They were terrible. So part of the reason is that the coordinating job the governing job adalah apa yang kita lakukan dengan baik. That so far AI’s have been trying or LLMs have been trying to copy. At best, I would say they’re trying to copy, but they are not doing a good, I don’t think they’ll ever get to where we get to because we’ll always be a step ahead. And I think how this ties into productivity and our conversation about embedded AI, I think that embedded AI is like the confident interns that if you ask a confident intern, how should I improve my productivity? The confident intern would say, great, here’s what I embedded. Here’s a way for you to improve the processing of your email. And the improvement in the processing of your email may have nothing to do with your overall productivity. It might even make it worse. Because the confident intern knows one improvement really well. And as Ray says, the confident intern does a search of five options and it comes up with the email. Optimize your email. That’ll make all the difference. But we at this higher level, at this coordinating level, would know better. And I think… We are. Need to recognize that what we’re doing or When we’re asking an AI, for example, tell me what productivity improvements I should make. I think it’s a bit like asking an AI what should our corporate strategy be? It’s a similar higher level coordinating question that The AI will never get to. And so be able to answer that question. We’ll always be ahead of it. It’ll never be smart enough. There’ll always be nuance and there’ll always be contradiction and there’ll always be a kind of intelligence, as Art said, that You do, you have, when you are on, I hate to use the word automatic because it carries us in the wrong direction, but when you’re the coordinator, you don’t pay attention to the detail at the bottom. So what Art said is true. You don’t know exactly what’s happening in the subcomponents. But that’s, O quê? Because your job is not in the subcomponents any longer. It used to be. Now your job is in this coordination intelligence layer level And that’s the part that we… I would say we’re confused as a society. Or as people were confused about the levels. And we don’t know how to speak about this intelligence that we possess, that the AIs will never possess. We don’t We haven’t coalesced around what this intelligence is and how it shows up in different fields like strategy or productivity. But now that we’re talking about it, I’m wondering if I could see it everywhere if I were to really look. I just never thought about it this way before.

Art Gelwix | 27:16

Yeah, I think that AI that you’re talking about that generates terrible strategy will be the first one promoted to CEO. When we look at… The role, and this is where I think we have to really start to frame within our own context, what are the roles that AI are completing for us or we’re placing in? We keep looking at, we hear the popular media, AI does all things for everything, for every way in every way. And that’s a terrible way to think about it. Take about it for writing, for example. Using AI as a researcher… As a research assistant is an excellent use of it. Using AI as an editor. After writing is an excellent way to use it. Using AI to automate the distribution process of your content is a great way to do it. But how often would you create a role for somebody to create the draft of your content. That’s a different thing. How much are you outsourcing your part in the process? And I’m talking about content creation there, but it could be just as something as simple as a quarterly business review presentation deck. We love that idea. I could push a button or tell AI, go generate the QBR deck for this meeting. That’s great. You have to have a factor. So if we look at the roles, we look at the understandings as to where this facilitates rather than replaces, I think it provides the context that we can apply it easily within the systems and the tools that we have. Unfortunately, so many of these tools that we have. Give us a big chat box, for lack of a better term. Say, go ask a question and it will ask all your data. Yeah, but that’s not where it’s really useful. So thinking about How would I ask a person to do this. And then go back and say, okay, now how do I ask this AI who’s really not that bright? But is really good at pattern recognition. How would I ask it to do.

Ray Sidney Smith | 29:31

It? I think it’s important for us to keep picking as like a level set. So if we have training data that is… What created the LLMs. That training data is basically the average quality of what was publicly available to the company that collected the data. So you are getting… Just basically garbage, right? Like it took garbage to create the LLMs because where is the most useful information? It’s behind paywalls. It is hidden away in communities that are gated. They’re not the public web content. While there’s very good amounts of it, there’s a high volume of it, which is why they used it to train on. It is not the best content for certain subjects. And so it, Remarkably, and I’m very bullish about where we’re going with regard to AIs, so far. There’s a point at which I think it’s going to break down with LLMs and we’ll have to find some other mechanism for doing what we’re doing. But right now, this is a step in the right direction, even with its fallibility. So don’t take me as not enjoying and appreciating what I have been looking for the past 20 years, which is this kind of tool. But it has been trained on, you know, Basically, the public web where people have opinions, they don’t necessarily always have facts. You have you this is being trained on conspiracy theories and on science, right? Like it is pulled all of those things together and is remarkably created. Highly competent simulations of natural language and thinking, right? It doesn’t think, but it does calculate in ways that makes it look like it’s thinking. And so that’s number one. We’ve got to be very mindful of the fact that we have to give it a lot of things, both samples. And templates so it can give us back good data because what it was trained on is the average, right? And Obviously, we think of averages as like bell curve, but in many cases, the average is Pretty stupid when we think about it in specific topics that haven’t been trained well, like strategy, right? That’s not going to be a topic where I would presume the LLM is going to be very good at because it doesn’t have the specificity in its training to give us the output that is going to be really truly tuned to its needs. As you create more utilizing AI and you don’t do the effort that I hear all of you talking about, which is crafting the final product into something that would actually look like you created it from scratch yourself. Guess what? When you do the next search of that It’s now pulling in that Subpar. Content. And now we’re dragging down that average in terms of the quality output. So every time you do this, you need to be very careful about what you are doing because the output then becomes part of the input for the next. And so you have this breakdown in the assembly line of quality. And so you need to have a little bit of a Kaizen approach here in terms of just continuous improvement on how you are producing. I fundamentally don’t think that this is removing effort from the humans, which means that the scale of what you can do does not increase that much. And so it goes back to what I talk about when we talk about Pareto principle. People bring up the Pareto principle. The Pareto principle breaks down very quickly for anyone who’s got their you-know-what together, right? The moment you have a task list and a calendar, you can generally be productive. All of a sudden, Pareto principle is no longer 80-20, right? You are now capable of getting so much more done with very basic organizational practices. So just keep that in mind. I think…

Francis Wade | 33:25

What’s muddying the waters? For what you’re seeing, Rui. Is that we don’t know It’s a little bit like So you’re the President of the United States, and you… Decide to and you’ve had mediocre, speechwriters. And instead you decided to go higher, the top. Harvard. Yale, whatever, writers, you find them somehow. You hire from the top 1% of the class and you bring them in as your speechwriters instead. And in the beginning, they are Amazing. Because they are producing a turn of phrase that you have only imagined possible. Delivering. And you tend to use more of what they tell you in the way that they tell you. Because you’re enamored. By what they’re telling you. And then what happens all the time is, You realize that they don’t have the whole picture. They are still confident interns, even though they are Harvard. And Yale-trained, confident interns. But your level of discernment goes up. In other words, you become more intelligent at managing them. And less enamored. You realize where you need to keep certain basic rules to yourself. You need to keep certain guidelines. You need to check their work. In different ways, but you still need to check their work. Before you used to check spelling, now you’re checking sources, for example, because they were more competent than their predecessors. But you’re intelligent. You’re a way of coordinating them. Inevitably goes up and inevitably increases so that you now relate to them differently than you did their predecessors. But Your role has no… Use the word that implied that your intelligence… Expands a level. You’re just as busy as you were before in the sense that your time is occupied and you’re thinking and making decisions. However, the things that occupy your time. Like spell checking and the way you spend your compute the things you’re thinking about. Or no, at a higher level. Because You don’t need to check that, the spelling anymore. I think this is inevitable and it’s confusing to people. People when they start to see the LLMs doing what they used to do better than they used to do it They get a little lost because they think that, replacement is about to happen. They don’t see their own intelligence as malleable, the Carol Dweck type of stuff. They don’t see that. They will become better coordinators because they have better interns. They might see the interns as threats, as some do, and say, I’m not going to use any of that LLM stuff. I’m not going to use any of that AI. And they just get rid of, you know, they just deny it altogether. They really ripped themselves off of the improvement that they could make. I think this applies in the productivity world as well.

Ray Sidney Smith | 36:26

So there are two ways in which folks are entering, utilizing usually in regard to personal productivity here, which is that They want the LLM chatbots to do things they currently aren’t doing. Because they perceive that it is beyond their capacity to do whether that’s time or capability or otherwise. That is not saving anyone time. So we can level set that if you weren’t doing it to begin with, it’s not saving you time. And so the second category is and are things that are augmenting your workflow and existing workflow. And is there to basically rate efficiency, or better work. Quality, or volume. And in usually both of those latter two pieces, quality and volume, there is more effort needed to get the quality and volume out even if you think it is saving you time, it is not. So it’s only on the first piece where you actually take a workflow and the AI actually speeds up the process that it becomes a time saver. So we’re thinking about this from very practically a personal productivity management perspective, we’re really talking about mostly effectiveness of איך משהו עשוי, נעשוי, בקבל שלא יעשו את המקום בצד שם, במיוחד או בפורם. אז אני חושב שרוצים לנסות לתחיל את הנושא של שהאי יש לנסות. But the most part is it’s not saving us time. And with a million interns, you need to do a lot of work onboarding them Every time I fully believe you should be creating documents, whether that’s a note taking system or Word documents or Google Docs with the appropriate source materials so that you can quickly and easily spin up. Context for them, saving your instructions, right? So having a prompt library where you’re saving prompts using text expansion software so you can quickly and easily get those things started. So you can speed up the process. That’s really good. And I think it’s a really good thing to do. But that is not going to in the large part, save money. That much time. It is going to make you look at the buffet and think, I want to eat everything. And we need to step back and learn how to say no to ourselves in the processes that we want to be higher quality. We want to get greater volume and better output. That is what we’re going for here. It’s not likely that I’m now going to take on five new roles or five new projects that otherwise I would only be able to manage one. Maybe you can go from one to two, but from one to 10 or one to 100, I think this is where corporate America has it wrong. And I think that certainly small businesses are not going to see this because you can’t hire. So you’re a small business owner. You have 10 employees, right? You’re likely not going to hire a chief engineer human resources officer. You’re likely not going to have a CTO, right? Those are roles that you just don’t have the money, literally, to be able to have on staff. So you may augment that with external services, right? You might hire an outside, a managed service provider, right? So an MSP for your IT services, you might have a payroll service, like a PEO or something else that is going to help you manage payroll and other kinds of human resources items. And then at some point, you’re going to invest in AI to basically fill in the gap between those two things. And I think that’s the appropriate kind of a model that we’re going to see invested in, but it’s not going to be, I’m going to have a AI CHRO or AI CTO, and it’s going to replace one who’s covering a 10,000 employee business. That’s just ridiculous. And so level setting where we’re entering in, how much it can do. I think all of those things are just practical ways in which we can both lead people and but manage the AI. As you noted, Francis, the reality that once you step into a management role, you’re no longer seeing how the meat is made. And if you’ve ever seen how a hot dog is made, you may never want to eat that hot dog again. And this is how it works with AI. If you understood the level of data that it’s looking at and the quality of the data that it’s looking at sometimes, you may be like, I think I should do this myself. And we just have to put our blinders on and give it really good source data, give it really good instructions, and then do the effort of fixing it until it is truly what you would want it to be. If you did it yourself… And it will make it a better output. And it may save you a little bit of time. But don’t count on it. Because you just really can’t. And I think it’ll get better. This is the hope, right, is that it will get better. I wanted to just touch on one additional item, which is that I think that there are really great pools out there that do very specific things. And I think sometimes people are afraid of augmenting their embedded AI by extending beyond those. So I think there’s two things to think about here. One is, if you can do the tool, in the existing framework then you should. Even if it gets you only to 80 to 90% of what you want. It’s still going to be better to use Copilot in Microsoft 365 than to go outside of it to something that is outside of the security provisioning of your corporation, outside the confines of the permissions and lacks the context. But Don’t be afraid to look at other options when it doesn’t meet that. Need. There’s going to be certain circumstances where you’re just going to need to have very specific tools that go beyond the existing framework. So I think the standard is look at the tools you have most often than not, The answer is Excel or Google Sheets. 99% of the time when someone says, we need it to do this, and I’m not quite sure we can quite do that. And I’m like, do you know how Excel and Google Sheets works? It can basically do… Pretty much anything. It’s a Swiss Army knife. They are the Swiss Army knives of the tool sets. You can pull in data, you can manipulate that data, and you can then output that data all through Excel, and Google Sheets. Generally, that is the answer. But if it cannot do that, Then you should look to another tool for doing that, like video generation. You’re really going to have to find a tool outside the Microsoft ecosystem if Clipchamp or another tool is not doing exactly the type of video generation you’re looking for from the LLMs. Okay, fine. You have to go beyond it. But for the most part, look inside the tool set. You probably don’t even know all of the tools that exist within your ecosystem. There are 500 plus products within the Google ecosystem. There’s 250 to 300 products within the 365 space. You have to kind of, you have to think outside your immediate tool set of the Microsoft Office suite being a limited set of tools to being a more expansive set. Final thoughts, gentlemen, before we wrap up this episode. Taking.

Art Gelwix | 43:47

The time to understand your underlying information. Is as valuable if not more valuable than learning a new feature that has been dropped on top of it. AI does not forgive early mistakes. So if you have made mistakes in the information you’ve gathered, how you’ve organized it, how you’ve structured it. It may cover those mistakes up. It may smooth them over. But those mistakes are still going to carry forward. You are responsible for your data. Not AI. I.

Francis Wade | 44:22

Think there’s going to be an ongoing challenge for us to capture, recognize and capture the new intelligence that we’re adding. Because. We’re adding the use of these confident interns. I think that’s the next frontier for us as knowledge workers.

Ray Sidney Smith | 44:38

So two things to close out. One, I think the most important shift is the gradual embedding of AI into our software environments where work already happens. So I think hopefully our conversation has underscored that element. It will make it less visible as a separate product in many ways, but that’s okay. Because I really believe that ultimately these should be commodity features in everything we use. And where we do not want to use it should not appear. And where we want to use it should. And so it should just augment our work by basically being a highly competent, feature within the tools that we’re already using. And then the second piece is that there is an opportunity to understand how AI is quietly enhancing the systems you already rely on and then learning to work alongside those capabilities without giving up your judgment and thinking. So while I believe in the opportunity of these tools, if you believe that cognitive demand is going to be reduced, then you are relinquishing part of your humanity to these tools and relinquishing That is a judgment call you get to make. I believe that I want to be able to tell my technology at any given time, any device about a calendar event, and it should be able to do the rest of the work. Find the location, put it in the calendar. I never want to use that cognitive effort again. I just don’t care about that, right? And that’s That is a choice I’ve made. I’m going to turn off my ability to have to calendar and schedule, but the tools can do that now highly competently, and it does not reduce my other thinking in my world. I think that it’ll be fine if I never have to calendar anything again in my life. It’s not going to happen right now, but it’s a guy can dream. Calendaring is a very big part of my day-to-day life, right? I’m always moving things around. Clients need to move things around. I want to put an event in. And if AI can do that to help save me the effort, I don’t mind. Internalizing that cognitive effort. But there are other things, my writing, my thinking processes around my work, the reading of material, all of those things are things that make me. And if I start giving them up, then I’m starting to give up part of my humanity. And I hope that everyone really understands that on a fundamental level. And even if you disagree with me, The practical aspect, but you’re going to do it anyway. At least you know what you’re giving up. And so I think that’s an important part of it. But with that, thank you, gentlemen. This has been a wonderful conversation, and we will continue in our next episode in this AI-powered professional series, and we’ll be moving along to talking about some vibe coding and seeing what folks are doing in the vibe coding space. See you then. If you have a question or comment about what we’ve discussed during this cast, please visit our episode page on productivity.net. They’re on the podcast website. At the bottom of the page, feel free to leave a comment or question. We read and respond to comments and questions there. As well, you’re invited to join our listeners group inside Personal Productivity Club, a digital community for personal productivity enthusiasts that I host, where you can interact with the ProductivityCast team directly. To join for free, visit productivitycast.net forward slash community, and you can get started there. By the way, to get to any ProductivityCast community, Episode fast, simply add the three digit episode number to the end of productivitycast.net forward slash. So episode 100 would be productivitycast.net forward slash one. Episode 102 would be productivitycast.net forward slash 101 and so on. On productivitycast.net on each episode page, you’ll find the show notes. So links to anything we’ve discussed are easily jumped to from there, along with text transcripts to read and download. If this is your first time with us, please consider adding us to your favorite podcast app. If you click on the subscribe tab on productivitycast.net, you’ll see the instructions to subscribe and or follow us and get episodes downloaded for free every time a new one comes out. And if you enjoyed spending time listening and learning with us today, it’d be a great help to us if you added a rating or review in Apple Podcasts or your podcast app if it has a rating and or review feature. Your compliments motivate us and they help us grow our personal productivity listening community. Thank you to those who have left reviews. We’ve seen them and appreciate all the feedback. Keep them coming. If you have a topic or question about personal productivity you’d like us to discuss on a future cast, please visit productivitycast.net forward slash contact. You can leave a voice recorded message or type a message into the message box and maybe we’ll use it as a future episode topic. I want to express my thanks to Augusto Pinaud, Francis Wade, and Art Gelwix for joining me here on ProductivityCast each week. You can learn more about them and their work by visiting productivity.net and visiting the About page. I’m Ray Sidney Smith, and on behalf of all of us here at ProductivityCast, here’s to your productive life.

Ray Sidney Smith | 49:44

That’s it for this episode of Productivity Cast, the weekly show about all things personal productivity, with your hosts, Ray Sidney-Smith and Augusto Pinaud, with Francis Wade and Art Gelwicks.