AI for Life: Tips, Tools, and Transformations with Celia Quillian

0:00We're seeing this huge eruption of agentic AI, which is where you ask it to go and do a task for you. It's all computer-based right now, but there is a world very soon, I think, where you can have an agent in some kind of robotic form go and accomplish a task for you around your house, kind of like Rosie from the Jetsons. Welcome to the Talking AI Podcast, where we talk AI with both experts in the field and early adopters. I'm your host, Matt Page, and we're here to demystify AI for you so you can get some value from it. Let's talk some AI.

0:34Adopting AI is hard for one core reason. Humans are lazy, we don't like change, and that's why building new and novel habits takes so long. But lucky for all of you today, we have Celia Quillian, Director of AI and Growth on Greenlight's innovation team and the author of AI for Life, a book that breaks down how anyone, regardless of background, can integrate AI into their daily routines. She's also been featured in publications like the Wall Street Journal, New York Times, Forbes, and Time. But welcome to the show, Celia. Thank you so much. Excited to be here. Yeah. And by the way, I love Greenlight.

1:09They're local to Atlanta. We're a local Atlanta company as well. My daughter feels like a little adult every time she's shopping or whatever. We can handle all the, making sure she's not spending it on something random, which is just a really cool, cool product, kind of an Atlanta darling here. So awesome to have you on. Thanks so much. Yeah, I love Greenlight too. I think it's so great that we're allowing to give more financial literacy to kids and teach them about smart spending early instead of later. Oh, exactly. Yeah, it's funny too. My daughter's like, she's eight years old, but Sephora is like this big thing now where all the kids go to.

1:47And, you know, she doesn't need makeup. So she's getting like lip gloss and things like that. But it's kind of funny. you want. But, you know, to start, I'm curious, tell us a little bit about your background, how you got into AI, both in as a career and on the creator side, which has been, I think, an interesting development as all this craziness has happened, especially since the ChatGPT moment a couple of years back. Yeah, sure. So I think at some point in 2021, maybe I joined some kind of a waitlist for ChatGPT. I think it was maybe or maybe it was Dolly. It was one of the two of those. And so when ChatGPT launched, I immediately joined it late 2022, started playing around with it and thought, this is so cool.

2:28You know, it was this constant magic moment. And I, you know, had the three days of late nights chatting with it to see what it was capable of. And it really quickly realized this new innovative technology is going to be everywhere very soon. And of course, we saw that ChatGPT, you know, broke records in terms of how many new users there were within the first month. So after a couple of months of playing around with it and chatting with people about it, I learned, oh, well, not everyone is adopting it as quickly as I would think they would in my circles. And I had always had an affinity for video production and storytelling.

3:08My background is actually in theater. So this is a great opportunity to share with the world what I'm learning as I learn it. So I started a little TikTok channel back in March 2023. And really rapidly, it took off. I was sharing tips about just, you know, every winning use cases of how I was using it to travel plan, to meal prep, to all these different things. And as I was sharing more about it, my channel grew and grew. And I just got even more into understanding what new tools are launching, what they're capable of. And through that, just kind of self-learning pattern, I developed a career in this AI space as well at Greenlight.

3:50So I was giving talks at Greenlight, teaching people how to adopt it. And now we're employing new technology at Greenlight as well. So that's kind of like the long and short event, just a passion-driven, curiosity-driven approach to AI. And here I am. It's funny too. I feel like there's back in the beginning, it was this like magical kind of moment, but it also wasn't very good either. I feel like there were some people that were like, oh, this thing's awful. And they just kind of wrote it off. And I think there's people like you that saw the potential. And I don't think any of us realized how fast it was going to accelerate and improve.

4:30But I think those that, and I think the thing too, I feel like so many people still feel like today, oh, I'm too behind now. But what I like to relate it to, we're at that moment where, you know, we got internet, but you still got to use the phone at the same time. You can't use the internet and the phone at the same time. There's so much potential. So I think any folks that are still early on and feel like it's gone too far, you should not take that approach by any means. Absolutely. Now's a better time to be getting into it, to your point, because they've improved so much. I mean, obviously there's still hallucinations and there's still some information that it generates that's not factually correct, but it's gotten a lot better at that.

5:11So the noticing that like, wow, this thing is not so good as it should be has definitely gone down. So people are trying it for the first time now and they can do so many things now, too. So multimodal capabilities within all of these free tools have made it that much more effective. Quick break in the pod. If you're listening to this podcast, chances are you've been thinking about how to actually use AI inside your business. And that's exactly why we built the AI Opportunity Finder. It's a free tool that helps you uncover high impact, tailored AI use cases based on your business, your goals, your pain points, and your industry.

5:47No fluff, no generic use cases, just real ideas that fit your business and the rank by ROI potential. It takes about three minutes to run and it's like having your own personal AI strategist for free. If you want to try it for free, check out the link in the show notes or go to hatchworks.com backslash AI dash opportunity dash finder. Yeah. And the stuff you can do for free is getting insane. But like we mentioned, it's inherently, I think humans, they don't like change, which makes this adoption phase different, especially when you have something that's completely different. Because I think people are used to a certain way of working and this is completely new and novel from that.

6:27But when it comes to like building new habits around AI, what, like, how should people approach that? Like, what is the, you know, stepping stones? How do you adopt the tools when there's like a new tool coming out every day? How would you recommend folks start to go down this journey? Yeah. If you have never really touched these tools or we've only kind of touched them in the capacity of maybe you've heard they can rewrite an email in an okay way. And that's the extent of what you approach it with. I think that the first thing I would recommend is find a free tool. There's a lot of them out there.

7:01Chat to T, Copilot, Gemini, Claude, Perplexity. All of these tools have a free version and just experiment with it. I think what I encourage people to do is when they are faced with a challenge or problem that they might typically turn to a Google search for, or just any kind of curiosity or challenge, anything that you need to normally write down or start a first draft on, try using an AI tool first and see how that feels. Just to have it, like try once a day to attempt that. And then I think the other side of that is, you know, approach it with curiosity. So if it doesn't give you the response that you would hope for immediately, recognize that these tools are conversational.

7:43They can call back to what you said previously and they can receive feedback much like an intern could. Right. So don't don't immediately write it off if the first response you get is not perfect. You can actually refine the response by continuing to chat with it. And as you do that, you'll start to learn the nuances of the language that you use and how it impacts the output. So you'll get better at getting a better response in that first go at the more you approach it. Yeah, there's several like rabbit holes. I want to go down there. I think the key thing you said, though, it's really it's building in that daily habit.

8:18I have a lot of people that I talk to. It's like, how do you know all these different tools and things? It's like, I just randomly am playing with it every day. The thing you mentioned is great. Like, just like consciously, don't just go to Google, start with ChatGPT, and you'll start to evolve how you approach talking to it. I think the one piece, there's a lot of folks that talk about this whole prompt engineering and this whole new job function that's been created. And then there was some backlash on that saying, oh, that won't be around. that won't be an important thing. And I've gone both sides on this to where it's like, oh yes, this is very important.

8:59Oh, this will just be a common skill, which I think it will be. But as I've gone deeper into it, I've realized the value of good prompting is very important. And I think like you've mentioned before, it comes down to good communication. It's almost like you're talking to a human. That's how you need to approach it. But I'm curious, like any tips or approaches when it comes to prompting? How do you think about prompting? You know, and there's, I think the difficult thing is there's tons of different scenarios where you may employ different tactics, but any just top level approaches you take when you're prompting with AI?

9:36Totally. Yeah. I think the times that I actually put more focus into prompting with AI is if I am building something for a repeatable task where it's using, you know, I want a consistent output regularly. So, for example, if I am building a customized wrapper, a customized version of ChatGPT, custom GPT, to create an engaging, compelling TikTok script, for example. There's something that I like to use as the kind of markdown format where you have a little hashtag to kind of give it some structure and format there where I say, you know, this is your job. This is your goal. This is your task. Include a lot of detail and specific context.

10:15this is what the example output should look like. But that is specific to those use cases where you are specifically trying to create repeatable tasks. Now I'm at the point, I joke, where when you start using ChatGPT you're using frameworks, right? You're like, okay, okay, gotta remember all of these specific things. One of the frameworks that I teach people is Sage Plus, which is specific context, action. So what is the action you're trying to get it to take? What is the goal that you need it to accomplish? and optionally, do you have an example? And the plus is you can always continue to prompt it.

10:49It doesn't all have to be in that first message, especially using it to chat. But once you move beyond the frameworks, eventually you just begin to talk with it like you're talking with a person. I mean, I rarely type my prompts out anymore unless it is for the purpose of building those repeatable tasks. Mostly I use the voice-to-text functionality to give it all the context it could possibly need and hope that it's communicated effectively enough and send it off and see what happens. I think for most people, that's going to become the approach to using these tools as well. But specific context is the first important thing that you can have.

11:28Yeah, I think we've all heard that content is king. I feel like context is king, is the new thing. And I totally agree. Like the more context you can give it, I think that's where a lot of people fail when they're working with these systems. They don't have enough context. They are very eager, ready to work. But if they don't have the context, they're going to try and answer the question. But, you know, just like that's where I think that's another flip people need to make. It's not human, but the more you interact like with it, like it is human, the better output you're going to get. And it's it's funny, like to that extent, like I was playing around with cursor, just building something random over the weekend.

12:06And I told it, I was like, OK, I got to go to my daddy daughter dance with my daughter. I didn't need to tell it that, but I like told it that I was like, OK, let's I'm going to stop here. And I gave it a reason why. And it's no no value added to it, I don't think. But I did it anyways. It's just because you start to like feel like you are working with a partner, in a sense, which is, I don't know, kind of a weird habit that started to materialize that I've noticed. Well, it's so true. I mean, it's the it's like when, you know, for a long time, people debated, oh, is saying please in your prompt going to give you a better result than not saying please?

12:39I don't know if everyone figured it out, but I do think that saying please and thank you in your prompt naturally gets your mindset in a place where you think and imagine that you were speaking with the person that's actually comprehending it back to you, which can make it easier when you're providing that context. Because if you think you're just talking with the computer, you're not going to say please and thank you. But if you say please and thank you, you're in the mindset of I'm communicating with an individual that has some level of expertise that may be able to help me with the problem.

13:08so it's like a subconscious flip or something and also like the the running joke too if you know they do take over one day they'll remember that yeah celia was very very nice and said thank you so she she can live yes and that and that kept me up to date with what was going on in his life and that was very nice yeah yeah exactly uh but another thing you mentioned too the multi-modality piece of it and i think this is another angle especially folks that are newer at it they don't realize all the things that it can do can't see and you mentioned interacting with voice and I've heard some folks doing this they'll just I think this is a great learning way like if you're on your commute or driving somewhere by yourself just start talking to it and have a conversation but curious like what where's the the value there what unlock do you see with the voice aspect of it versus traditional kind of writing and prompting and whatnot right so for me I actually use these so So the advanced voice mode where you can actually, it talks back to you like a person.

14:07It's fun and entertaining. And to your point, great on a long commute where you're toward someone's top of your head. You don't want to grab your phone and search or chat with ChatGPT, obviously. And it can kind of also get you in the mindset of communicating with an entity, right? My favorite feature is just there's a little microphone in almost all of these tools now that you tap it and it voice records for you what you're saying. And then it transcribes it perfectly. I think that that saves me so much time. I'm a pretty fast typer, but being able to speak aloud, everything that's going on, just kind of do a brain dump and then send it off to chat GPT and let it consolidate it, summarize it, do whatever I would like it to do with it, saves me a ton of time.

14:52I mean, I get to the point where I record for five or six minutes saying, leaving a meeting. We just had a meeting. We talked about a lot of things. Here's the detail of what this person said. this person said, these are the key takeaways. I'm just brain dumping what just happened. And it acts like my secretary. At the end of all of that, when I'm jumping around and say, oh, and another thing I forgot about this thing at the beginning of the meeting. At the end, I say, okay, all of that, I need you to summarize into nice, concise, bulleted notes with key takeaways at the end of this meeting. Send it off.

15:24And I've got a one pager that I could send out to everybody within five minutes versus within me, you know, reformatting it itself writing it out um so that's a huge unlock um and yeah that's that is the primary way that i approach prompting it now because it just saves a little fraction of extra time yeah in in those fractions they build up over time in a big way and maybe the last thing on prompting so we obviously build software so we're using it in the development process but i'm curious like the parallels here so i was using it over the weekend and i've taken this journey which i think a lot of people have where you say hey go build me this thing and then you get in these like crazy rabbit hole nests and things and it just blows up right but i've started very intentionally and i'll start over with like oh one right or some reasoning model pick your reasoning model and very similar to what you just said i'll kind of give it the overview the high level thing I want it to do.

16:23And then I'll have it give me very detailed requirements. But before I do that, I say, you know, do you have any questions for me? And then go through this iterative process where I get these very nice, tight requirements. Then I go over to cursor or pick, you know, whatever coding agent you want. And then versus saying, okay, now go build this thing. I have this great list of requirements. I then talk to it and say, Hey, you know, I'm looking to build this thing. I have these requirements. Let me know what questions you have. Because technically it's new in context for this thing. And technically you could talk to a one in cursor, but I, I did it over in chat GPT.

16:58And then I said, okay, you know, and then once it says context for the thing, let's work together to define how we should approach building this thing in one order. And then it's logically thinking through, okay, well, here's, you know, the architecture I want to build based on these requirements. This is the order we should do it in. So we don't screw things up later. And what you've done is you've given AI context for what you want to build so you can work together. And it's just amazing how much better it is. And I think I've seen you talk about a similar thing. It's almost the, I think you called it like priming AI.

17:34But curiously, and for any folks, before I forget, that are interested on the coding side, there's an amazing, I guess called a blog post, this guy that wrote this thing, how to stop saying FU cursor. And he kind of talks through this whole process, very similar to this, where you kind of got to flip how you interact and think with it. And there's a couple of folks that are really getting deep into this. We have a framework called generative driven development where we're doing the same thing. But talk me through your thoughts on like the priming of AI that I've heard you talk about before. Yeah, yeah.

18:07I talk about pre-prompting it and also to your point that asking it, am I forgetting anything? Is there anything else you need to know? both of those be great to give you the additional context that you don't know you need to have right so um the pre-prompting element the priming is you know if you're trying to i think i use in my example a linkedin post you're trying to write a great linkedin post and trying to get started on linkedin talking about your career talking about whatever um maybe you don't know how to write a great linkedin post you don't know what hooks people in you don't know what um compels them to like or comment or whatever.

18:41You could ask ChatGPT in the chat to begin with, can you create me a guide to writing the most effective LinkedIn posts to drive engagement from this kind of audience and effectively grow my professional brand? It will create a guide for you and it'll have a lot of tips and tricks of how to format it and everything else. And then you say, okay, great. Can you write me a LinkedIn post about this now? And you just go into your, you know, it's the topic you want. And then it's pre-primed on this is the context that I need to follow for this particular kind of LinkedIn post they're going for. And then to your point, be asking it to ask you questions.

19:18I talk about this a lot in my book. It gives you that extra layer of, OK, there's probably something I've forgotten that might be valuable for it to know. And, you know, you can even say, like, for your use case, acting as the software engineer, ask me about any technical work that I might have forgotten. acting as the ideal user. Ask me any questions about what I might be considering from your perspective, you know? And all of that can provide additional information that you can use to create something better. That is, I think, a key thing that I think people that are hesitant about adopting AI or scared about AI forget, is that, and I truly believe this, that AI is not going to take everyone's jobs, but someone that knows how to effectively collaborate with AI is going to be better off than someone that does not because effectively using AI is about that collaborative approach to it.

20:12It's not, okay, I'm going to send you one thing, you send it back and we're good. You've got to continue to build with it. You've got to continue to collaborate and brainstorm and bounce off each other, quote unquote. Yeah. Yeah, that's great. And I think you mentioned another thing too. It's kind of priming it with the persona that it should have, but you can also change its persona too. I've done that before where, okay, you're an architect now, now you're a software engineer, or now you're a strategist. Now you're a product person. Now you're a copywriter. So, um, that also just gives it again, context for how it should act.

20:46Uh, and you mentioned that one point I've gone back and forth in my own mind on this. Like, is it going to take jobs? Is it going to enhance jobs? I've gotten to this point where I feel like those that adopt it will be far and ahead, the winners in this. So go adopt it. The deeper it gets, I do feel like it is going to disrupt some jobs. But I do feel like also it's going to create new opportunities, jobs that we don't even know, that don't exist today. Oh, absolutely. Is the other thing. But I don't know. Maybe give me your take. Pick your time frame. five, 10, 50 years. Like what does it look like?

21:29How are we interacting with it? What does the world look like? As this to the topic in your book, it's not just pervasive in work, but like in your life, it becomes native, like we use our phone or whatnot. Yeah. Oh man. Even going out more than a year at this point is just like science fiction. I mean, already we're seeing this, this huge eruption of agentic AI, which is where you ask it to go and do a task for you. It's all computer-based right now, but there is a world very soon, I think, where you can have an agent in some kind of robotic form go and accomplish a task for you around your house.

22:04Kind of like Rosie from the Jetsons. I don't think that's out of the possibility. Certainly in the next few years, maybe even the next 10. I think that the workplace is going to look very different. I think that using generative AI tools is going to be akin to how we approach using Google or email today very quickly, like in the next couple of years. I think right now a lot of businesses are in a place where they're still trying to figure out what their policies are generally around AI usage. So we're not quite there yet, but I think within the next five years, certainly most of them will have figured it out because it would be foolish not to.

22:42There's just too much productivity gains to be had. I do think it will disrupt some industries. I do think that it will also help a lot of industries be more efficient. I mean, customer service is one that people think, you know, that's the first one to go. But there's still a lot of customer service roles that I think are going to be necessary because they have some like highly sensitive information they're dealing with. You know, you don't pass information through an AI to a certain degree, but there are going to be parameters where these are the situations where you can use generated AI to help with tickets.

23:13It's the situation. Can't. But yeah, I mean, it's going to sound like something of sci-fi if I go too far into the future. Yeah, it really is. You wonder how quick we get there, if we get there, what it all manifests into. And you mentioned the customer service. I was playing with this this weekend when I mentioned I was building something, this tool called Bland AI. And there's lots of these like 11 labs and whatnot. The voice tools, they do something cool, though. there's this uh kind of architecture they have on the back end where you think of traditional ivr you have to define like each you know thing that may happen if this happens then you say this then that this structure you're giving it more so guidance for scenarios and then it has the llm attached so it kind of knows you know what direction it'll go so it's really interesting to see how you know these, these new architectures and approaches are starting to, to manifest.

24:11But you mentioned in the workplace and I've seen organizations that are on the bleeding edge of this and they're adopting it and I've seen some that are just way behind and I'm just curious, what's, what's an approach you think that, that works maybe how is green light approaching it? Cause I think sometimes there's this, um, analysis paralysis that happens within And especially large organizations where they almost can't get out of their own way to start adopting these tools because it is so new, disruptive, unknown. But how should organizations be thinking about this? Yeah. I mean, at Greenlight, obviously, we have a dedicated team, innovation team.

24:54In addition to all the projects we're working on, we are the ones kind of at the forefront of experimenting with new tools. while our company is still figuring out where we abide with our, you know, generative AI policy and how strict we are and where, you know, whose law to use it for what, we can kind of go in and test these things out. So having a dedicated tool that is sourcing and looking at different things and keeping a pulse on what's changing is very valuable. I do think to your point, analysis paralysis is a real thing, especially when you have all of these, you know, innovative AI, multi-purpose companies that offer an enterprise package and all of them are competing with each other.

25:31So one day, one is the best, the next day, another is the best. I think at a certain point you just kind of have to recognize, okay, we'll pick one for this year at least. And it's going to maybe be the worst one day, but it'll be the next two days from now. Yeah. So that you don't totally lose your momentum. And then I think, you know, effective training of your of your employee base is incredibly important too because there are obviously risks to using generative ai tools and if people think that it is magic and it's always correct you're going to end up in a pickle because obviously these tools can hallucinate even the best of the models can with more advanced mathematics or anything else and like understanding what data can and cannot be uploaded into these tools from a privacy standpoint from a compliance standpoint is incredibly important.