← The Amazing Intern - Turning Your A.I. Curiosity Into Confidence
The Amazing Intern - Turning Your A.I. Curiosity Into Confidence
Ep 63 - Intro To Deep Research & A Customer GPT
00:19:57
In this solo episode of The ChatGPT Experiment, host Cary Weston shares personal updates, answers a listener question, and dives into two powerful tools within ChatGPT: Deep Research and a Custom GPT for Ideal Customers. Cary explores how these tools...
Loading summary
Transcript1 lines
- [00:00]
Kari Weston
Foreign. Hey there. Welcome to the Chat GPT Experiment. This is the podcast designed to help curious folks better understand just what this whole ChatGPT thing is and how to find some practical and valuable ways to use it in your own life. My name is Kari Weston. I'm your host. Hey, hey. First time listeners, welcome, repeat listeners, welcome back. Appreciate you joining me today. I got a couple things here. First of all, how you doing? Forgot to ask, how you doing? Doing good. I'm finally getting over some bug virus, whatever this thing is. It's been holding rent in my head here for the last month. I was at the gym Monday and trainer finally said, hey, look at that, you can breathe again. Just walking in the door was getting out of breath. So, you know, it's interesting. I tell my kids, you know, the only thing you want to be when you're sick is not sick, right? And the only thing you want to be when you're injured is not injured. And when we're not, we forget how much we might take for granted some of the things that we have when we can actually function as we normally do. So good to be back. Good to have breath again. Good to be able to do things that I normally, quite honestly, it's good to be able to walk up the stairs, not have to bend over and catch my breath. So hope you're doing well here. Got a few things for you. Thought I'd go to the mailbag first and then I'm going to get into real quickly. I've been experimenting with the deep research tool. Have you played with this, the deep research tool in ChatGPT? I'll tell you what I've seen and learned and give you an example of what I've done with it. I'm early into this one, but I thought I'd share it with you because I think it's kind of cool. And then I've got a custom GPT for marketing folks, for business folks. I guess nonprofits too would be valuable if you've got customers. Basically, this custom GPT will be helpful to give you an objective voice, to have kind of a focus group with you 247 so you can run ideas, run copy, get input, get some clarity, that kind of stuff from your ideal customers. Okay. And so for the customer GPT, I'll link out to the website Chat GPT experiment. There's a document there that you can download. It's all free, but the document has the two part instructions to it and more on that in a minute. But let's go to the mailbag first see, Ian down in Texas says, liked your episode that you recorded from your truck. I know you use an iPhone. I find that when I record, I don't get the transcript. It has my audio recording, but I don't get the transcript. Or it takes a really long time for the transcript. What did you do? Okay, so, yeah, I find the same thing. So when I record to my iPhone, I use an audio recorder tool, and it has an audio file on my iPhone, right? So once I have that audio file, I can treat it like anything else in my phone, like a video or a picture or whatnot. So actually what I do is I exported it. And because I use Apple on my desktop and my phone, I can just wirelessly send it over. But you could email it, text it, however you want to do, but via Bluetooth. I sent the audio phone, the audio file over to my computer. And then I actually use for both this recording that I'm doing now and all of the transcript stuff that I use, I use a program called descript D, E, S, C R, I, P T. And I brought that audio file in, and what it does is it creates a transcript from it. And here's the cool thing. If I were to have two people or three people involved in that conversation, descript will listen to that audio file, create a transcript, and while it's doing it, it'll say, hey, listen, I recognize multiple voice patterns here. Can you tell me who they are? And it'll play a snippet of the first, the second, and the third, and you give it a name. And so it'll go through the transcript for you, identify those voice trends, and label it for you. Right. So then you've got a record of who said what. So hope that helps. Yeah, I find that I do find the same thing. The transcription sometimes doesn't come as soon as we want, and so I use other tools for it. So I hope that was helpful. Hey, listen, if you've got a question, you can find me@chatgptexperiment.com There's a form there. Just. Just shoot me a note. Love to hear back from you. Hey, the custom GPT for customers. Oh, yeah, Deep Research. Want to go there first? Deep Research. Have you tried this out as you. I have the paid version, so I think it's probably a paid version utility. But at the bottom of the conversation tool, right, the screen, there's a place called Deep Research. And just in playing with it, I found a couple things that I want to share with you. One of the things that Chat GPT has acknowledged is it doesn't know everything, right? Which is very cool because as I've shared with you before, Chat GPT is programmed not to say I don't know. It's always programmed to give you an answer. And if you've played this at all, you know that sometimes the answer isn't right, or sometimes it's made up, or sometimes it has variations of the truth and variations of fantasy. So one of the things that Deep Research is doing really well is it's connecting Chat GPT to the outside world of resources, okay. And it's basically recognizing that it doesn't know everything, right. And what it'll do for you. And it. It's not instantaneous. So we'll share this with you and makes sense because it's taking the cue from you and it's using its powers to go out and connect to resources that will help its get. Help it get the information that you need. So a couple things that I'm finding. Number one, the better the input, and this is a generic tip, right, that I'm always saying, but the better the input, the better the request. The more detail you give it, the better the output's going to be. So again, let's go back to our definition of AI being an amazing intern or amazing instrument rather than artificial intelligence. If you think of it as an amazing intern, you've got to give it some clarity. Think of an intern that has tremendous capabilities, but it doesn't have the context, the details, what success looks like, right? The background from you. And so the more detail you can give it, including, by the way, why you're asking it to do research, the better you can give Deep Research that information, the better off it's going to be, number one. Number two, it does take a while. It will literally process for. I've had it go up to 10 or 15 minutes. And it does have a progress bar. It says that's doing all of these things. What I haven't done in my experiment is, is leave ChatGPT while it's doing that. So what I cannot tell you is, can you be multitasking in ChatGPT and doing something else while it's doing that? I don't know. My workaround so far has been to just fire up a different tab in my Browser and bring ChatGPT up and then keep using it that way. But I cannot tell you, if you leave that conversation while you're still in the same window of ChatGPT, if it will continue to process. So if you Know the answer to that. Shoot me. Known. There's only one way of knowing, and that's to test it. And then if it's in the research and you bail on it and it doesn't, then you've lost your research. Right. I guess you could fire another one up, but I just don't know the answer to that question. So, number one, the level of detail matters, although it's very good with limited, but like everything else, the better you give it, the better output you're going to get. Number two, it's going to take a while. And by take a while, 10, 15 minutes sometimes. And then number three, it's very thorough. And not only does it show you the sources that it's gathering, it actually lists them in the document or in the output with links. So you can actually reference the external resources that it used to create whatever it is that you ask them. And I'll give you an example of what I've done here in a second. It cite, it cites them in the work that it gives you. And then it has actually has a list at the end of found of resources that it use and links to it. So it is going out. Think of it as a research assistant. It's going out. It's looking at various places online based on the parameters and the requests that you have, the goals that you have. It's bringing those sources back, it's using it and processing it in the analysis that it's creating for you, it's referencing those links and then it's sharing with you at the end the links and the resources that it used for you to verify. Right. So again, I'm new to this, I'm young to this. I've had limited reasons to use it at the moment, but I'm hearing good things from others that are doing it. Our friend Marcus Sheridan went to buy some equipment for his business, did some deep research on equipment. What's being sold, what the prices are being sold at the moment, what's being listed, what the features are, what the trends are on that particular equipment and pricing and features and models, and use that to negotiate with a salesperson to get a better deal because he was better armed. What he found was he was better armed and better informed perhaps, than the person sitting across the desk from him that was selling him the product. And that's just one level of thing that I've read out there from. I can imagine the $10,000 putt. The $10,000 golf putt. Someone shared with me once that if I shared with you that I'd like you to come in next week and I'm gonna give you one chance to make a putt. A golf putt in my office or a golf putt on a golf course or whatever it might be, I'm gonna give you one chance to make it. I'm gonna give you $10,000 if you make the putt. Most people would probably practice before they go to make the putt, right? And the analogy was that many of us have $10,000 putt opportunities every day in terms of meetings or customers or pitches or whatever it is that we might be doing in our business. But we don't practice. We just show up, right, and take the putt. So the $10,000 putt analogy applies here to deep research. If you have a meeting or if you have a sales pitch or a collab, something that you should be most prepared for, and it's important and matters to you, and you think that preparation would help you. This deep research tool I think could be something that would be of value to you. I know I've done this in pitching companies. I want to know about them. I want to know who they are, what their competitors are, what their trends are. And I haven't used deep research to do that. I've used ChatGPT to do that. And it has provided me some valuable insights. And I could imagine the level of sophistication in additional research that this tool would allow would make that even better. So deep research a minute. So let me share with you. I A couple episodes back, I told you I did some workshops for contractors for hardscape and landscape contractors on ChatGPT for business. And I'm going to put this. It won't be on the Apple, it'll be linked on the Apple version of or your podcast software version. It'll be on the website because I can't put documents on the platform. So if you go to chatgptexperiment.com under episodes, look for this one and I'll put the document there and I'll show you what I asked for and what it gave me. And you can see the resources that were cite, the references that were cited and how it does it. But in preparation for that, I said, listen, I know this is my audience. These are landscapers and hardscapers and contractors in Maine. I want to understand the industry so I can make my presentation relatable to them. Could you tell me what the top three trends are in Maine for hardscape and the hardscape and landscape industry? And actually As I say that out loud, I didn't say Maine. I said, can you just tell me the top three trends? It came back and it said, do you have a. It was really smart, right? Because it said, do you have a geographical reference in mind? Because the materials that are used in outdoor landscaping and hardscaping differ depending on the climate, right? I said, yeah, that's good. So I'm worried about Maine. So it came back and said, great, we're going to just focus on Maine and then you'll see in the document. It took about 15 minutes for me to get the document that I posted on ChatGPT experiment. But you can see what it gave me. And it gave me a good level of detail, some analysis, the sources that it came from, and again, that reference list. And it made me prepared. I was able to take that information and work it into my presentation to make it relatable to what's happening. Right. So really, really cool. So deep research. So chatgptexperiment.com go to this episode and you'll see the document you can take a look at. And I included the prompt. And this was a very skinny prompt, by the way. I gave you all of the details here that are on that document as far as what I asked for because it kind. It resonated. But again, in other instances, the more detail you give it, the better the output is going to be. Okay. The last thing I'll share with you today, this is something I've done with a few customers lately. It's becoming terrifically helpful and I thought I'd share. And again, on this podcast episode on the website chatgptexperiment.com I'm putting a document that has two instructions to do what it is that I'm sharing with you. So back in episode 44, this is for marketing, this is for business folks. And again, anyone with a customer, so nonprofits, this would apply to you too, because you have donors and members and sponsors and whatnot. Just as businesses help customers. The episode 44 was how to. It was an overview of how to use ChatGPT to define attributes of your best customers. Best. Right. Fit customers. Right. So this is taking that a step further. So what I've done is I've created two sets of instructions for you. Two exercises. One is a deeper dive into using ChatGPT to get not only the attributes. And this is. And this is the cool part. Not only the attributes of your best customers, but the emotional positions, the empathetic positions, the worries, the frustrations, thinking like Them, right? So taking on the perspective them. So it's not just a bucket list of things. This is going to enable you to take that one step further, which is give me the list, but also give me the emotional components of what they think, what they feel, and then you're going to use that list. And this is all in the instructions. If you use the paid version, you can create custom GPTs. You're going to take that list and you're going to literally copy and paste the output into ChatGPT to create a custom GPT. And the details are in this document. So you can create an objective bot that you can use over and over again to act as the voice and perspective of that customer, right? Or that member or that donor or that sponsor or whatever it might be. So step by step, instructions on how to get real deep and granular into your best right fit customers. The attributes, the questions, the frustrations, the positions, what they're looking for and what they're thinking, what that sales call might, might contain for questions and areas, all that stuff. It's going to give you a really long, detailed list and then you can use that list, create a custom GPT in chatgpt for your ideal customer. And then here's what you're going to end up when you're done is you're going to have a tool that you can rely on, ask questions to and have it answer and give you feedback through the perspective of that best right fit customer. Here's how I've used it. I create a lot of copy. I work with clients to create both web copy and email. I have a client that does. We create videos every other week. We have a filming and he creates videos for his websites and for his outbound marketing efforts. And what we do is we take the perspective, we use the bot, the customer bot that we fed all of this information into and we outline our thoughts to create the article or the email or the script. And we say, could you read this and let me know, Would you be confused? Is it clear? Would it excite you? Would it create action? Do you have any questions? That kind of thing? Are we talking over your head? Are we talking about things you don't care about? Could we make it better? Could we make it punchier, whatever it might be, right? And it literally comes back and gives us information as if we're using a focus group of our best customers and it gives us some really good feedback on how we can make it better. Is it perfect? Of course it's not perfect, but is it the best thing that you can come up with right now to actually replace talking to real customers. Sure, nothing's going to replace talking to real customers, but imagine if you could model all of the attributes and questions and emotional positions and fears and anxieties and questions and then use it over and over and over again. You could pitch ideas, you could say, how make my website better? How can I make this email better? I just want to use this subject line or this subject line. Which one gets you excited? Which one would get you to move? I want to use this in my marketing message. I want to create a workshop around this issue. What is the things that you. There's all kinds of things, right, that you can come up with. So again, chatgpt experiment.com if you want to take the attribute list of your best customers and layer on top the emotional positions, the problems, questions, needs, goals, frustrations, all those things, and then create a your own custom GPT that would replicate and act as that customer as you're using ChatGPT. Those instructions are right there in that document. Right. And as always, if you use this, I'd love to hear both from deep research and from this custom customer GPT exercise. Give me some feedback, let me know if you're using it, how are you using it, what you found, was it helpful? You know, are there any recommendations? Would you. Or does this stem more questions? And is it possible to do this or have you done that? I'd love to hear from you. Okay, so that's it for today. Hope you're doing well again. I do some trainings and workshops. I've got a couple more coming up from listeners. Thank you for doing, checking in and looking forward to those. ChatGPT experiment.com you can find backlog of the episodes, some articles, the workshops and trainings that I do. Reach out to me if you have any questions and of course, the resources that we talked about for today's show. Okay. And as always, the number one component to you being successful with ChatGPT is your own curiosity. Okay, so as we wind this episode down, going to say do stay curious and we'll talk soon. Bye.