AI Agents for Designers

Thanks for listening to this episode of "Designing Success: From Study to Studio"! Connect with me on social media for more business tips, and a real look behind the scenes of my own practicing design business.

Grab more insights and updates:

Follow me on Instagram: https://instagram.com/oleander_and_finch
Like Oleander & Finch on Facebook:https://www.facebook.com/oleanderandfinch

For more FREE resources, templates, guides and information, visit the Designer Resource Hub on my website ; https://oleanderandfinch.com/

Ready to take your interior design business to the next level? Check out my online course, "The Framework," designed to provide you with everything they don’t teach you in design school and to give you high touch mentorship essential to having a successful new business in the industry. Check it out now and start designing YOUR own success
THE FRAMEWORK ( now open) https://www.oleanderandfinch.com/the-framework-for-emerging-designers/

Remember to subscribe to the podcast and leave a review. Your feedback helps me continue providing valuable content to aspiring interior designers. Stay tuned for more episodes filled with actionable insights and inspiring conversations.
...

Transcript

0:29

Welcome to Designing Success from Study to Studio. I'm your host Rhiannon Lee, founder of the Oleander and Finch Design Studio. I've lived the transformation from study to studio and then stripped it bare and wrote down the framework so you don't have to overthink it. In this podcast, you could expect real talk with industry friends, community connection, and actionable tips to help you conquer whatever's holding you back. Now, let's get designing your own success. Today I wanna talk about something that's about to change the way creative businesses operate. AI agents. Some of you may have heard of genic ai. Some of you may know nothing about AI agents. Some of you might think I'm an AI agent. Some of you might have built an AI agent. I don't know where we're at in terms of how prolifically we're using. AI and which LLMs or large language models we're using. I generally talk to chat DPT. I also use Gemini, Claude, which is Anthropic and perplexity. So if anyone is actually curious about the sorts of things that I do, I use them for different. Reasons and for different outcomes. And I'd be happy to do a podcast episode on those. And why I am always really conscious and I'd love your feedback. Come and talk to me over at Oleander and under Finch in my dms on Instagram, um, or reply. There's a, a bit at the top of the show notes, it says, text me and tell me what you thought of today's episode. I would really love your opinion on whether you want more of. These, I'm always really conscious. I don't want to flood ai, ai, just absolutely everything. AI into my podcast. It's sort of really always been a mixture of strategy, marketing, business design, friendship, interviews, conversations, and not just robot talk. So I try to pepper them in where I can. But if you want more, come tell me. I'll give the people what they want. I just don't want to nerd out on you. All of the time, because when I'm in a nerd hole, I'm in a nerd hole. Like I'll stay there forever. I will tell you and talk and share and never gatekeep, and like, let's, let's have lessons. Let's build an agent together, live on a podcast. I'm so happy to do that. But today I just want to talk about AI agents because I do really believe that they are changing the way that businesses operate. They've changed the way that mines operated, and I don't want you to just be like, oh, this is gonna be like a tech. Nerd out and I'm like, please don't hang up on me right now. I promise that I'm got not about robots taking over your job. It's about finally having the power to create a tool, a agent that can take on boring, repetitive process, things that really, it's not your zone of genius. You shouldn't be doing it. You could be out at a fabric house, you could be drinking your pumpkin spice latte, perusing the showroom floor at Cocoa Republic, having the best day of your life because everything's being taken care of. I think there's a real resistance. I spoke to someone at a, after a keynote recently a couple of months ago who was talking about, I was talking about AI agents with her, and she said to me, oh yeah, but I wouldn't trust anything to write an email. And, I think it's disingenuous and I really understood what she was trying to say. But I was also like, okay, Sarah, you keep typing random replies to requests in your inbox that don't matter. You know, things like. Things I just need to reply to on a day to day, my son's doctor asking a single question. Someone asking me, did you do X, Y, and Z? Or someone saying, can you remind me again what we did on this project at this time, it's all in my Google Drive. The agent could literally go and source the answer and reply to the person, and I didn't know anything about it. And I'm bloody here for that. I don't care if there's an M dash here or there as long, you know? And, and I'm very big on being the last person to run my eyes over and sign off on something before anything is deployed by AI in my business. So, just so you know, if you get a, an email from me or a message from me and you're like, Hmm, that looks AI generated, I have read every word of it. Whether I've decided to highly edit it, like I will always audit my return on investment. Is this worth me editing the crap out of it and making, putting my own spelling mistakes as I usually would have them into this email? Or does this look better, more professional, get the point across and have I signed off on sending it in this manner, in which case I've just hit sent. And even if it looks a bit ai, I've been part of it. I just didn't create it from scratch. So I understood what she was talking about. Her name was not Sarah, by the way. All the Sarah's listening, you don't need to be like, oh my God, was that me? I totally got her point, but I also would refute it a little bit. I think we've gotten beyond that as a. As a technology, I think it's so good what we can get our agents to do with all the right training so that they understand and that they are writing decent emails, right? We can teach them to write the way that we write, we can give them all of our past sent history for the last five years and be like, learn my phrasing, learn how I write and get them to do that. You just have to give them the right rules, and this is what's often missing is vague prompts. Vague rules, vague agents. You know, you've got to be specific. It is a robot. It needs recipes, it just needs the formula. And this is what I do when I'm doing technology development for other businesses, creating these custom tools for businesses. That's what I do. If you have a business out there and you're thinking, oh my gosh, wouldn't it be great if we had a, an email assistant? Had all that stuff and that list, you know, that wrote like us, how would we design that? I could do it for you, just get in touch. That's beside the point today. But I think we, we need to finally step up to the light and say, this is here. It's gonna change the way we do everything. And it's gonna help you spend so much more time being creative and so much less time chasing your tail with admin. And who doesn't want that? So if you resonated with that last sentence that you're an interior designer who's just, chasing quotes, sorting files, retyping the same brief, retyping the same email, doing similar repeated tasks 17 times in one week. This episode is for you. All right, let's kick off with what is an AI agent in plain English, I just, I want you to understand the difference. As I was joking before, it's not me. I'm not a secret agent for ai, or it's not a person. If you've used chat GPT, you've met what's called A GPT, so that is a. Where you type something in, it gives you an answer. It's responsive, it's conversational, but you do a lot of the prompting, right? So you may have one of my gpt and they're skilled to do particular tasks you might work every week. For example, with Otis, who, who understands the algorithm, he writes your Instagram captions, helps you bring lots of keywords, lots of personality, and beat the algorithm, get great results, right? That is A GPT. Now there's still absolutely a place for GPT. There's a whole bunch of things that A GPT can do where it prompts you to think and then you give your answers and you elevate the outputs and you make it better and better, and you're working together. Robot and man is one. It's great. But you know, it. It still requires that conversation. So you wouldn't just want to set up an automatic agent called Otis that just writes and publishes Instagram captions for you every week, because I feel like there would be none of you in that. You wouldn't be choosing the topics, you wouldn't be brainstorming, you wouldn't be coming back with feedback. So, that is your GPT. But an AI agent is the very next step. It doesn't wait for you to tell it what to do every time. It just knows it's wrong. It knows its recipe. It has access to tools or data, and it takes action for you. It's not just a chat bot, it's kind of like you have hired someone new into your business, you've given them your standard procedure, your standard operating procedure. You've given them the exact steps that it needs to take, and you've said, Hey, when action A happens, someone sends an email, Read the email. Find the necessary information that you need across my Google Drive. Draft a reply, send it to me for review, and I'll send it out I'll. Greenlight it, or maybe you're like, send it to me for review. Wait for me to thumbs up the email or flag it or mark it as whatever. There's a step there and then you are good to go. You send it to the client. I don't know, I'm just making this up, but it's an assistant who runs tasks in the background while you focus on your work. So you do your thing. It does. Its. So I'm gonna give you a simple use case for an interior designer, and that's, and it's something really practical. Imagine you have a digital assistant inside your studio called the materials agent. You train at one time, you tell it where your supplier list is. So I might say I have a trade supplier list of 160 trusted trade partners that I have. My login details saved in there like I've got, I've got a trade program, et cetera, et cetera. All saved in notion, so I would tell it where my supplier list is. I would tell it what kind of materials I specify them most often. Now I could either just tell it that or I could have. Gotten CHATT PT to connect to my Google Drive and tell me what I use most often, which I probably would'cause I'm a bit of a cheater like that. If I can find a way where it can extract the data for me, I don't have to think, and then I could add to that things that I could think of, and then you could tell it how you like things formatted. In your fff e schedule or how you'd like the information delivered to you. Is it a table, is it a analysis report? Like what is it? From then on, every time you upload a new client concept or product brief, the agent automatically searches your trusted suppliers online. So I'm gonna put a brief in to this agent. I'm looking for, I need, the client wants. This is what we're doing. It'll go, it'll search all of my trusted suppliers online, find matching finishes or furniture within the price range that I've given it, and then pull that data into a table, for example, with the link so that I could go off and have a look. Oh. Thanks for that agent. Thanks for that materials agent. You know, I would give it a far better name definitely, but you know, if you've got the materials agent, thanks. Materials agent. Now I'm all set. I can take what you've gathered for me and it's a starting point. Now. I can go over, I can go to the A BI website, go and have a look. Yes, I agree. I want this mixer or no, I don't. Or I, yeah, it could just help you. That's a three step process. It doesn't change. Go to notion search. The websites of all my trade suppliers match the brief. Present it to me. I'll take it from there. No searching 10 websites, no endless email trails. Like, we could build this would be great. I don't want to. Dilute the power of curation and the power of sourcing. I just think that there's some really cool applications to get us started. Sometimes we overcomplicate things and things are as simple as they are. How many brushed copper taps are there out there? You know, if the first three come to me from a BI, the blue space and force it strong and I know that these are trusted suppliers of mine and I really like them, then I can choose one copper tap out of those three and feel confident without having to open a whole bunch of different websites. So I could get my pre-filled schedule ready for approval and not spend my time refining. Like I, that's not how I do my FFDs. I definitely do them one click from Project Studio, but I could ask it to give me all the images so that I could upload that information if I didn't have them in my library already. It's just, there's so much magic that an AI agent could bring. That's one example, but it could be so much more simple if somebody. Books a discovery call. I'll take all the information over the call and I upload the transcript and you open the CRM that you populate all of the client details into the CRM home address, property address name, phone number, email, everything from the transcript of the call where I collected that information. Boom, done. Rinse and repeat over and over. There's so many different ways that we can simplify and think about how could I have these little agents build so that whenever I. Click on the widget, click on the button. That action activates. It's not it. The core functionality of an AI agent is to remove the repetition that slows your creativity down. Okay, this is a big one.'cause they know that you're sitting there going, yeah, all good for you. You know how to build agents, but where can you start even without coding, like not knowing anything. Clearly, it's easy for me to say I can run Python, I can run JSON coding. I, know what an API is. I know how to connect things, but I assume many of you do not. So this is the good news about Chat T'S Agent Builder. You don't need code to build your own AI agent from scratch. You just need to start by documenting what you do. And I cannot be strong enough on this. This is why we have the. SOP Master Library in the framework express 132 documented processes so that we can start to train ai. You have to know what step comes next, what, what are the steps, simple steps so that you can train the robot to execute the steps. If you yourself don't even know the process, you can stop listening now. Sorry to be rude. AI agents are not happening for you. Like you need to stop and you need to document what you do. That is the first step. Now that could. Look a lot simpler now that we have ai, right? Like you might have something somewhere, or you might just spend. An hour on your own Zoom call talking to your own face. No one's in that Zoom call and just going, when I do this, I do that just so you've got the transcript right? Or like an audio recording and just like map it all out verbally. Then I would send an invoice. Then the client would let me know that, or I would receive the payment. I reconcile it in zero. Then I go over and send them a, I open up their client portal, I load them into the C. Like what are the steps? You have to know this stuff. You have to be documenting it. cause the second that you do, you can share it with AI and get it to do the thing on your behalf. All right, so this is how we're gonna use AI agent. You're gonna pick one repetitive task, something you do at least once a week, sending, oh, I probably don't wanna like updating a client file or. Even something to do with your emails, right? Write down the exact steps that you take. Pretend that you're teaching a VA or you've just hired someone. You're like, okay, well I need to actually tell her how to update the client file. What do I do? Then I do this, then I do that. That's it. You've done that. You've got the steps. Then you turn it into a role description, so. You can work with AI to do this. You're an AI assistant that sources materials from my supplier list and formats them into my FFF and e template. That's the starting brief. For your AI agent, for example, or you are my admin assistant, you will be updating client files. You'll need to know the information that needs updating the area to apply the information to, and you'll need to be able to AC access Trello cards and amend them or something like, I'm just making this episode go along, but you need a brief for the a i agent. And this is why this is worth paying attention to. Designers who are starting to experiment now are the ones who will build real operational advantage later. Like I feel like if you are someone who's like, oh, why don't I go watch a YouTube clip on no code AI Agent builder, there's like. Thousands of them already eight hours after I looked at Open AI's Dev Day or Development day where they launched Agent Builder, there was already YouTube tutorials on how to build one end to end. So go learn it. It's not, it's not hard. It just takes time, you know? But it gives you smarter studios. You are going to have the operational win at the other end. If you know how to do this. I mean, you could pay someone to do it. Obviously you can get development assistance. You can talk to me, you could talk to. Thousands of developers on the internet. But how cool if there's someone in your studio who knows how to make one.'cause then if you have a new idea of something you wanna automate, you don't have to constantly hire someone. You can actually go, all right, let me see if I can work out how to. Pull this through Agent Builder. It's not perfect. It's not gonna replace Zapier sticky tape for the internet, or NANN. There are so many different reasons why I am not gonna go into them on this podcast episode. There's a lot of tech Bros came out as soon as it was released and we're like, it's gonna kill NAN and Zapier, and it's this, and it's that. It's not, it's glitchy, it's basic. It doesn't have the ability to MCP properly. That doesn't mean anything to you probably right now, but watch YouTube clips or you know, have a little Google and,/and get into the nerd world, the good thing about this agent builder is you don't have to be a nerd to build it. It's not techie. It's actually about efficiency. And if you rise above the fear of, oh, it could be really hard. Yeah. Or it could be really easy'cause it's supposed to be built so that. Every business owner can access it. So this is my challenge for the week. Pick a process that slows you down and just start mapping out that process and think about what data or documents it touches along the way. Like what does it need to know about? That's your first step towards your own AI agent. And then watch the YouTube clip and see if you can work this out with AI Agent Builder. Now it's not just in church, bt it's in a different website. It's like AI agent builder slash platforms or something. I'll find the website. But. Honestly, if you can treat AI as a team member, not a threat, you're gonna run circles around designers who don't. I think this episode hopefully has just gotten you thinking you could send me a dm, we can have a chat. I can talk to you more about this. I'm very conscious that this episode, if I'm too specific, we'll be irrelevant to people within about three weeks. So, um, I know the rapid. Growth and change that happens in ai. So every time I do an AI based episode, I am keeping it kind of wildly vague. The things that you could do. What an AI agent does is it has button number one is start, and then it's just basically like when the button is pressed, that triggers the next movement. Now the next logical movement after start is guardrails. And guardrails are protections on the internet. So you build your agent to start and then to double check that the person is not a cyber threat, that they are a person and to lock down their secure information. So if this person talks to you about their credit card details or about the blah blah, blah, you're not to share anything. So there's a cool little guardrail part that OpenAI have included, which absolutely love.'cause it means that you can't accidentally build an easy hackable agent that people can get into your, um, design processes and into your business. So that's good guardrails. Good. Make sure you include that. And then it basically is like if, then what if this happens, what happens if this action is taken, what do I do? You just kind of build it like a flow chart. You just say, I want you to do this and then do this, do this, and then this, and you build it out until the last step is sort of then deliver this. And then you are finished for the day until someone presses the start button again. So you're building a workflow, teaching it. It's rinse and repeat roll and pattern behind the scenes, what data it can access, what it could connect to, like it could connect to Canva or connect to a Slack or another app. So I could have one that's a customer chat bot that sits inside, let's say like an agent.*/That's like a, what would rri and say kind of agent that sits in Slack for all my framework is like that's where they all. Chat. Personally, I'll probably still show up for them'cause I feel like that's what I promised to do and I really, I love our chats in there and then they're not, it's not too taxing. But I could have one that, you know, even triaged and answered it if it was a logical, easy answer to find outside, inside of the f the data, like the framework, the course. If the answer was in the course, then the chat bot answers it. And if it the answer is nuanced or needs my opinion, then I answer it. In fact, I might build that one this week because I think that sounds fabulous. So, yeah, if this episode got you thinking, if you're interested in AI agents, let's chat. Just email me, chat to me in the dms, come and talk to me about what you would design like. Sourcing proposals, onboarding. I'm really curious to know what you guys are going for. Are we keeping it really simple? Are we triaging emails and just tidying up all that crap? We wish we'd unsubscribe to? Like, what are we building? Uh, options are endless. It's not as hard as you think, but I will in all, I will always just give you my honest opinion and I will tell you that I think you will struggle if you have zero development experience. I do think that they've made it very. Accessible but it's not necessarily ready for everybody just yet, but I think you could learn a little bit, like, I think there's enough YouTube and other places that you could go to learn should you want to set aside some serious time. Or I think now that you know that there's a possibility you could get an automation built for you, I think it's definitely an area that's worth investing for the new year. Saying like, let's build three agents, or let's, let's scope it into a developer to get three things built for our business that we can rinse and repeat use and not really be part of anymore. And to start slowly building an army of little agents. I love this for us. Okay. I hope that has demystified AI agents and why they're different to GPTs. Why there's room in the world for both of those things and why they're both equally as important. Yeah. One is just activating something with not a lot of nuance, and the other is unpacking and conversationally getting to a specific desired outcome. Okay. I'll chat to you next week. Bye for now. That wraps up another episode of Designing Success from Study to Studio. Thanks for lending me your ears. Remember, progress over perfection is the key. If you've found value in today's episode, go ahead and hit subscribe or share it with a friend. Your feedback means so much to me and it helps me improve, but it also helps this podcast reach more emerging and evolving designer. Just like you for your daily dose of design business tips, and to get a closer look at what goes on behind the scenes, follow at Oleander and Finch on Instagram. You'll find tons of resources available at www.oleanderandfinch.com to support you on your journey. Remember, this is your path, your vision, your future, and your business. Now let's get out there and start designing your success.