Interior Designers: There’s A Reason Why Your AI is Underwhelming
AI doesn’t understand you and your business.
In this episode of the Interior DesignHer podcast, we're talking about the real reason AI keeps giving you generic, underwhelming results; and the reasons why it has almost nothing to do with prompt engineering.
You've got a difficult client. You finally decide to let AI draft the email. Everyone told you this was going to be the easy part. And the result was fine. Not great. Fine.
So you spend 20 minutes going back and forth, adding detail, rephrasing things. The result gets better. You rewrite half of it anyway. And somewhere in that process you think “what is this actually doing for me?”
Maybe you blamed yourself. Maybe you blamed the AI. Maybe you just went quiet, because admitting your AI results are underwhelming — when everyone else seems to be getting exactly what they want — feels like admitting you're a BlackBerry in an iPhone world.
If that's you, this episode is for you.
The gap between what you were promised and what you got is real. You didn't miss something obvious. And the fix isn't a better prompt. Let's get into it.
Apple - Youtube Music - Spotify - Youtube
-
[00:00.0]
You've got a client who complains about just about everything, and you finally decided to let AI help you draft an email for them. That was supposed to be the easy part. The tech bros told us that AI would take over all of this boring stuff in our jobs and in our real lives, setting us free to be creative and follow our passions.
[00:21.5]
Another bunch of tech bros said AI would be a matrix, style end of the world dystopian apocalypse. And for about six months, we all freaked out and made a bunch of stupid AI meme videos. But when you asked AI to help you write that email to that difficult client, the result was it was fine.
[00:41.3]
Fine. So you spent some time going back and forth with the AI, adding some detail, rephrasing things, giving the AI more context, and the result got a little bit better. So you tried again. Twenty minutes later, you had something you could actually use, but you'd rewritten half of it yourself anyway.
[01:01.0]
And somewhere during this whole process with the AI, you thought to yourself, what is this actually doing for me? How is AI helping? Maybe you blamed yourself. Everyone else seems to get more out of AI. Maybe you're just not technical enough.
[01:17.7]
Maybe you blame the AI. It's overhyped. It's not ready for what interior designers actually do. And maybe you just went quiet because admitting the results that you're getting from AI are underwhelming. When everybody else seems to be getting exactly what they want from AI feels like admitting that you're a BlackBerry in an iPhone world.
[01:42.4]
So you pulled back, you used AI less or stopped using it altogether. If that's you, this video is for you. The thing is, you were right to feel what you felt. The results that most people get from AI are underwhelming.
[02:00.6]
AI influencers aren't going to tell you that this isn't in your head. The gap between what you were promised and what you got is real. You didn't miss something obvious. And none of the tech bros told you that. The way most people use AI, including most interior designers, skips three of the four things that, determine whether you get something useful or something generic.
[02:25.5]
And the one thing everyone focuses on, it's the least important of those four things. So why does AI keep missing? Here's the reason. And once you see it, you won't be able to unsee it. Think of it this way.
[02:40.9]
When you first sit down with a potential interior design client in their home and they describe their dream living room or dream kitchen to you. You don't start immediately pulling samples and cranking up the autocad. You ask that homeowner a bunch of very specific questions so that you can understand what they really want and what their dreams are, what they feel, what they actually need, what is possible, what is affordable.
[03:08.8]
Without that information, your interior designs won't be all that they could be. Unfortunately for us users of AI, us humans, AI doesn't ask those questions. And if you don't tell it everything, and I mean everything up front, AI reverts back to its training data and it gives guesses.
[03:30.0]
And this is a big problem with AI. It knows just about everything, but it doesn't know you. And if you're asking it to help you with something that requires that it understands you and your business, it's going to give you underwhelming results unless you make it all the way to the end of this video.
[03:51.4]
If you can hang in with me, I promise that by the end, you will know exactly what AI needs from you to get it, to stop guessing and, and to actually give you outputs that will make a big difference in your business. Okay, here we go. There are four things that determine the quality of output that AI, gives you.
[04:10.8]
And I'm going to walk you through them from least important to most important, because I want you to feel exactly where the real leverage is. The least important thing, but the most talked about is prompt engineering. Don't get me wrong, prompt engineering is important.
[04:29.1]
A better crafted ask does help. But here's the problem. Prompt engineering is the simplest and least important thing you need to get right if you want to get the most from AI. I like to think of those four things as layers in a cake.
[04:44.8]
Prompt engineering is the icing. Icing is great, but if you don't have those other layers to your cake, all you've got is a great big bowl of icing, which is yummy, but it's not a cake. So what are the other three layers? Layer three is context, otherwise known as what AI needs to know about you and your world.
[05:08.3]
Context is everything that AI needs to know about your specific situation before it can give you a specific answer. AI needs to really, really know you, your business, your design aesthetic, your process, your clients, your market, et cetera, et cetera, et cetera.
[05:28.2]
Without that detailed information, AI fills in the gaps with assumptions from its training data and it gives you outputs designed for this generic interior designer avatar that it's built from this training data.
[05:44.2]
And you are not a generic interior design avatar. Layer two is intent, otherwise known as what you actually want to happen, not the task, the outcome. What changes if your interaction with AI actually works.
[06:02.3]
Intent is the layer almost nobody talks about. Intent is the difference between write me a follow up email to a stalling client and I need this client to stop second guessing the tile selection and approve it before my lead time kills my schedule, derails the project, makes everybody freak out and cost me a bunch of money.
[06:21.9]
Those are completely different asks, same task, completely different intent. AI can only aim at what you tell it to aim at. Layer number one is specification, otherwise known as what done actually looks like.
[06:39.5]
Specification is the clearest picture of the output you need from AI. Telling your AI that you need an email or a business report that is professional is just generic AI business speak. the very least, you need to tell your AI that you want something like a three paragraph report, conversational, no bullet points, ends with a clear next step.
[07:05.7]
Reads like a person, not a company. And when I say that's the bare minimum, I mean that's the bare minimum. Whoever told you that all you need to do is talk to AI like another human being doesn't know what they're talking about. Without this layer of specification, AI decides what done looks like.
[07:26.7]
And then we're right back to generic assumptions and underwhelming outputs. Here's the thing, if you're already getting decent results from AI, you're that small part of the population that does well with AI. There's a reason why you're getting decent results.
[07:43.6]
Every time you go back and forth refining your ask of AI, you're actually filling in these layers for the AI. You're adding context, clarifying your intent, getting more specific about what done looks like. It works.
[08:00.4]
AI will make fewer assumptions. The problem is this manual way of doing it is slow and you're starting from scratch every single time. But what if you did it the other way around? Populated those layers before you asked anything.
[08:19.3]
Let me show you what that looks like with two real interior designers. These are real sessions I had recently. I just changed their names. The first session is focused on improving AI image generation for an interior designer named Sarah who is using AI or trying to use AI to create initial concept renders for her clients.
[08:41.6]
And she has been underwhelmed with the results she's been getting. Sarah has a new client interested in a, kitchen renovation. She has pictures of the existing kitchen. She had an initial discussion with the client and had some concepts bouncing around in her brain.
[08:59.4]
Sarah wanted to use, an AI Image generator. In this case, she was using Nano Banana, but it could have been Mid Journey or any of the ones that are more specifically marketed towards interior designers to help her generate some initial concept renders before she actually cracks open her autocad and gets down to work.
[09:19.8]
Sarah's client was having a hard time visualizing the words that Sarah was telling her. So, as they say, a, picture is worth a thousand words. And Sarah wanted to use an AI image generator to generate that image to make it easier on the client to understand.
[09:36.3]
Unfortunately, Sarah's prompts to her AI was generating subpar images and actually confusing the situation. So, as I said, for this job, Sarah asked nanobanana to take a look at the existing kitchen photo she had and replace the existing white shaker cabinets for light oak square inset doors.
[10:02.4]
The response from AI was an image that gave her those square inset doors, but they looked more like a yellow painted surface than actual light oak. It's not the end of the world, but it's not what she was hoping for. When I sat down with Sarah and we worked over this example, I identified a bunch of gaps in her approach, to talking with AI.
[10:26.0]
Here are three of them. Light oak is a description, a reference image. If she had attached a picture of what light oak cabinets look like, that's a specification. Sarah did have a reference image on her phone of the exact Oak tone she wanted.
[10:45.3]
But instead she had described oak in words rather than showing the image to AI. That's a specification gap. Sarah knew exactly what she meant, but she hadn't told the AI exactly what she meant.
[11:02.9]
To get a better output, Sarah should have shared that image. Another issue was the placement of hardware on the cabinets. Sarah knew exactly where she wanted the cabinet hardware placed, she told me, two and a half inches from the corner opposite the hinges.
[11:20.2]
Bottom corner on upper cabinets, top corner on the base centered on the drawers. Unfortunately, she never told Nano Banana that she expected it to know. Or maybe she just didn't think about it. The thing is, Nano Banana has no clue, and that's a giant context gap.
[11:37.8]
Layer three of our AI cake. Sarah forgot that the professional knowledge that lives inside her head doesn't exist in the brain of nanobanana or chatgpt or Claude or Midjourney or Gemini or any AI.
[11:55.1]
The third gap that I identified for Sarah was where she wrote magazine quality lighting. In her mind, Sarah imagined her AI generated image looking like one ripped out of a fancy interior design magazine.
[12:10.7]
But she forgot to share that specific intent with nanobanana. Even if we just look at the lighting piece of this, prompt Sarah wanted in her head. Dramatic editorial, magazine quality image, deep shadows, high contrast, all that stuff.
[12:31.4]
But she told the AI magazine quality, assuming that the AI would have a reference to what a magazine quality image would look like. But it didn't. AI heard something like, nice lighting, same words, different understanding, subpar images.
[12:53.4]
That is a massive intent gap. She knew the feeling she was after, but hadn't translated it into something that the AI could actually act upon. Just those three little mistakes. A specification gap, the light oak cabinet context gap, hardware placement, and an intent gap.
[13:12.1]
How she wanted the image lit resulted in a subpar image, an annoyed interior designer and an unimpressed client. It would have been better if she had never used the AI image generation software at all. To be quite honest, when I helped Sarah eliminate all of her AI prompting gaps, the prompt immediately improved.
[13:35.0]
When her prompt improved, the AI generated image improved drastically. That's what AI is good at. Give it a better prompt, it will give you a better result. But we weren't done there. I asked Sarah to dive a little deeper. I really wanted to understand the project and what she wanted from these images.
[13:54.4]
So we dove deeper and we uncovered other specification, context, intent gaps beyond those three. The images got better and better and better. Sarah got the result from AI that she wanted. She just needed to address those four layers.
[14:11.1]
Okay, let's move on to our second interior designer. Our second interior designer is a kitchen and bath specialist in my hometown of Toronto. She posts organic content regularly on Instagram, mostly professionally shot images of her amazing projects, professional photography.
[14:30.1]
She also posts in project reels and images to show the process. These. These are images and videos she takes on her iPhone. She gets tons of likes and shares and has over 7,000 followers and regularly grows that follower count each month. But she doesn't get as many DMs as she wants.
[14:48.9]
She gets very few website visits and almost zero clients that she can tie directly to her Instagram. But before we dive into her AI session, I want to introduce you to the AI tool that I use with both of these designers to help them get better prompts and better results from their AI.
[15:08.6]
I call it AI Sherpa, and I built this thing for myself. It's not a prompt template, it's not a course, it's not a community. It's a set of instructions that you can paste into any AI chatbot, Claude, ChatGPT, Gemini, or a specific AI tool like Nanobanana or Midjourney that turns it from a quick answering machine into a question asking partner.
[15:36.5]
Instead of giving you an immediate answer, it asks you questions, hunting for those nitty gritty details about what you actually want, what you need, your specific situation and what a good result looks like. The four layers that we've been talking about, it doesn't produce anything for you until it has everything it needs to give you a very high quality result.
[15:59.2]
To ensure that we were comparing apples to apples, our kitchen and bath designer opened up, a session in Claude using my AI Sherpa tool. And she used the exact same opening line that she had used previously with her regular version of Claude.
[16:17.4]
My Instagram isn't getting any traction. I post consistently but nothing is working. Not a good prompt. But to compare apples to apples. We're going to use that exact same prompt with my AI Sherpa tool. When she typed that line into Claude, here's what she got back from Claude.
[16:35.1]
Post reels consistently write captions with a strong hook. Use location hashtags like Toronto interior design. Aim for three to five posts a week at a call to action. She'd heard all of this before, probably tried most of it. Nothing in that advice knew she was a kitchen and bath specialist at the very least.
[16:54.0]
Nothing new. Her referral business was already working for her. Nothing new who her ideal client actually was. Generic advice for a generic interior design avatar that Claude had in based on its training data.
[17:10.3]
AI Sherpa, responded differently. Instead of advice, it asked a question. When you say you want clients from Instagram, what does that actually look like? Are you trying to get people to reach out directly or build visibility that makes referrals happen faster? Our designer had to stop and think about that for a second.
[17:29.1]
Instead of giving her an immediate answer, Claude, armed with AI Sherpa, was asking her questions like a business coach would. So she replied, I want them to reach out directly with DM's website visits. Cold homeowners eventually becoming warm leads.
[17:44.7]
AI Sherpa responded to this by saying, so referrals are working. Instagram isn't converting. When someone lands on your feed, what do they see? Our designer replied, mostly pictures of my kitchen and bath renovations. AI Sherpa, asked another question and when you caption those posts, what are you writing?
[18:02.9]
Our designer replied something like amazing new kitchen design in the annex neighborhood of Toronto. For me, this is where the whole thing starts to get interesting. AI Sherpa asked Her. What do you know about that amazing new kitchen? What can you share about that amazing new kitchen that the photo can't show?
[18:22.5]
Our designer wasn't sure what that meant. AI Sherpa, explained instead of writing amazing new kitchen in the annex neighborhood of Toronto, you could be writing, this client had a 10 foot galley kitchen and two kids under 5. She needed an island, but everyone told her it wouldn't fit.
[18:40.0]
Here's how we made it work. Same photo. And it gives potential clients a completely different reason to stop and think, hey, that's my problem too. And the light bulb went off in the brain of our interior designer. Our designer got it.
[18:56.2]
AI Sherpa, asked her to share a story from this kitchen project. What was the client's situation going in? What made it hard? What did you figure out? What was the story? Our designer replied that she actually had a significant problem on that kitchen job.
[19:13.0]
As part of the design, they had to open up a wall. And when they did, they found black mold. Remediation, ate up a big chunk of their budget. Every selection suddenly had to work twice as hard. And yet it became one of her favorite projects.
[19:29.3]
The constraints had actually made her work be more creative. And just like that, we had the perfect story to go along with the perfect picture for the perfect kitchen. Now let me show you what AI Sherpa actually surfaced in that conversation.
[19:46.0]
Because this is where the layer cake comes back. The first thing AI Sherpa, found was an intent gap. Our designer came in hot, saying she wanted traction. But traction isn't intent, it's a symptom. When AI Sherpa asked what traction actually looked like for her specifically, she had to think what she really wanted was Toronto homeowners planning a luxury kitchen or bathroom renovation.
[20:13.2]
People willing to pay for quality and experience. Not Instagram followers, Not likes, not saves, not traction. A very specific client with a very specific mindset and a very specific budget. The thing is, this is universal.
[20:29.4]
Every interior designer watching this video right now is looking for their ideal client. Not followers, not Traction. A specific person with a specific budget who values what you do as an interior designer. That intent had never been shared in any AI conversation she'd ever had before.
[20:50.4]
So when her regular version of Claude gave gave her advice about reels and hashtags, it was aimed at growing a, following on Instagram, not attracting her ideal client. Avatar. If your AI doesn't know your intent, it's going to give you generic answers to a question you didn't actually ask for.
[21:09.5]
A generic interior designer avatar that it is made up in its own AI silicon mind. The second thing AI Sherpa surfaced was a massive context gap. Our designer's feed was full of beautiful finished photography, stunning kitchens, immaculate bathrooms.
[21:27.3]
And here's the thing. At the luxury end of the market, beautiful is assumed. A homeowner about to spend some serious money on a renovation takes it for granted that the end result is going to look amazing. What they're actually worried about is is everything else.
[21:43.3]
A home renovation is stressful. Mistakes happen. The value is in how those mistakes get handled. Whether their designer communicates when things get difficult, whether the experience matches the price tag. Her Instagram feed was showing the destination.
[22:00.0]
It was completely silent about the journey. And the journey is exactly what that client is buying. The mold story isn't just a good caption. It's proof of this. It's context. Renovations sometimes go sideways. What homeowners want is a designer who stays calm, gets creative, and delivers that beautiful result.
[22:21.5]
That's the context a luxury interior design client needs before they'll pick up the phone. It was all sitting in her head the whole time. Nobody human or AI had ever asked her to say it out loud. Now that Claude was armed with all of that intent and context, it built a pretty darn good caption to go along with her existing pretty picture.
[22:44.0]
Here's what Claude, armed with AI Sherpa said. She called me about her galley kitchen in a midtown Toronto. Semi tight, dark, not enough storage. When we opened the walls, we found black mold. Remediation took a big piece of the budget.
[22:59.3]
And suddenly every design choice and selection had to work twice as hard. But isn't that when interesting design decisions often happen? The kitchen we ended up with is one of my favorites. Not despite the constraints, but because of them. If you're planning a renovation and worried about what you might find, that's exactly the conversation to have before you start.
[23:22.7]
Same designer, same kitchen, completely different story. Put yourself in the shoes of a Toronto homeowner and. And you read that, how are you going to feel versus just looking at a picture of this beautiful kitchen? That's not the end of this story.
[23:39.4]
We haven't even discussed our, kitchen and bath designer's specification gap. She came in wanting help with a caption. That's it. One caption for one post and then another caption for another post. Each one separate what she didn't know is that we could take everything that that conversation had surfaced and build it into a system.
[24:00.5]
In this case, it would be a set of instructions that she could put in a Claude project built on her intent, her context, her voice that she can use. Every single time she picks up her phone to post something on Instagram, she describes the project. Claude, armed with these new instructions, will build the story in her voice aimed at her ideal client, showing the journey behind the destination.
[24:25.3]
Not once. Every single time. She had no idea that was possible when she started talking to me. That's what specification does. When you get it right, it doesn't just answer the question you asked, it builds something that answers it forever. These two stories are exactly why I get so excited and so annoyed with AI.
[24:47.7]
AI has the potential to do so much for us in our businesses, yet without specification, intent, context, and as a result, proper prompting, it is so darn underwhelming. And that's why I built the AI Sherpa originally for myself, for my own business, my own life.
[25:05.8]
It's not a better prompt or a set of instructions or an online course. It's a guided conversation that forces your AI to ask you the right questions about what you actually want to happen, your specific situation, and what a good result looks like.
[25:21.5]
By the time you get to the actual ask the actual prompt, AI has everything it needs to give you something truly amazing. Which is exactly what those tech bros promised us at the start. If you want to try AI Sherpa for yourself, it's not ready yet, but I've got a wait list going and I'll attach a link to the description of this, video.
[25:47.1]
No more people pleasing AI nonsense. The moment your AI starts asking you questions, it really is going to change how you think about AI and what you're going to get from AI. Promise.
Key Takeaways
AI doesn't ask questions before answering. If you don't fill in what it needs to know upfront, it falls back on assumptions built for a generic interior designer, not you.
Prompt engineering is the least important of the four layers. It's the icing. Without Context, Intent, and Specification underneath it, all you have is a bowl of icing.
Context is everything AI needs to know about your specific world before it can give you a specific answer. Your design aesthetic, your clients, your market, your process, your voice. Without it, AI fills the gaps with assumptions built for a generic interior designer avatar that has nothing to do with you or your business.
Intent is not the task. "Write me a follow-up email" is a task. "I need this client to approve the tile selection before my lead time kills the project" is intent. Same ask, completely different output.
Specification tells AI what "done" looks like. Without it, AI decides — and its definition of done is generic by default.
At the luxury end of the market, beautiful is assumed. What your ideal client is actually buying is the journey — how you handle the hard stuff, not just the finished kitchen.
Conclusions
What Changes When You Fill In the Layers
Interior designers who get real results from AI aren't using better prompts. They're giving AI what it needs before they ask anything. Context about their clients, their market, their aesthetic. Intent about what they actually need to happen — not just the task, but the outcome. Specification about what a good result actually looks like. When AI has all of that, the prompt almost doesn't matter.
What This Looks Like in Practice
Start with a single interaction you're already having with AI regularly — a proposal paragraph, a client email, an Instagram caption. Before you type the ask, write down three things: what AI needs to know about your situation, what you actually need to happen as a result of this interaction, and what the output should look like when it's right. That's the minimum. It's slower the first time. The second time it's faster. Eventually it becomes the system.
The four-layer approach is what Doug built AI Sherpa around — a guided conversation tool that asks interior designers the right questions before producing anything, so AI has what it needs to give genuinely useful output. The waitlist is open.
The Bigger Picture
The gap between what AI was supposed to do for your business and what it's actually doing isn't a technology gap. It's an information gap. AI is doing exactly what it can with what it's been given. When you change what you give it, the output changes. That's not a promise about AI getting better. It's a description of how it already works — when you know what it needs.
Join the Interior DesignHer Community
If you enjoyed this guide and podcast, you're going to love all the other business education resources we're bringing to interior designers.
As interior designers ourselves, we created Interior DesignHer to bring top-notch business education to interior designers. We especially have a soft-spot for interior design solopreneurs & micro-businesses.
Subscribe now so you never miss our latest business guidance, guest interviews, and insider strategies specifically for interior designers.
