Faraz (3:25)

Thank you, Sam. It's a pleasure to be here.

Sam (3:27)

It's great to be here in Dubai.

Interviewer (3:28)

I appreciate you sitting down with me. I'm just curious. The most important question to me is, you know, what's driving you to become curious about using AI as a practitioner and a lighting designer?

Faraz (3:42)

Okay, this story starts from 2022, actually, when AI hit everywhere. ChatGPT was there. Before ChatGPT, there was Midjourney, the visual software that I mostly use, that most of my team uses. During that time, I started seeing these images flood the social media. These amazing, stunning images, which were more of artistic in nature. They were not realistic, they were more towards the artsy side. But it drew my attention. I could see that it had the potential to do something. It had the potential to do something real good, not in terms of just the artsy styles, but in reality as well, to aid what our workflows, to aid what we are doing as creatives. And the way it kind of developed so fast, it evolved so fast. Very soon I knew that I would be using these to do my creative things as well. And pretty much, I think within about a year or so, we were doing that. So as midjourney advanced, as their technology advanced, as the data sets, they learned more and more. They could render out more photorealistic outputs, and that's where we got it into our workflows. I knew that, see, when I have to present a project, I do sketches, I do renders to present the initial story. And then my, you know, my. My sort of workflow for presenting a concept has always been tied to, you know, connecting it to a metaphor at first. So in order to do that previously, I used to do sketches, I used to do renders, and somehow I used to connect those renders to the actual project images. But then with AI, I could. The things that I was doing in maybe about a week earlier, I could do that maybe in about three, four hours. And that's how it started. And it was effectively communicating, you know, that visual sequence of metaphor to move the image to the real project renders. It was sequentially communicating it in a very effective way.

Interviewer (6:01)

How important is prompting in all of this?

Faraz (6:05)

Well, prompting, as they say, is the. It's something which is sort of a soft kill of the new era, new age. So because prompting is where it starts, prompting is what makes that machine understand what you're trying to say. So the thing with them is that it's not just the simplistic language. If you go above and beyond and try to tie in a little bit of poetic elements, a little bit of, you know, literature elements, and, you know, metaphorical elements, even metaphysical elements, it reacts it in a very different way, in a very unique way, and that's what makes it different. So prompting is where it starts, actually. That's where you develop your base, and then you iterate and iterate and iterate till you finally get your desired image. Sometimes you get it in the first two or three attempts. Sometimes it takes about 20 atoms for the machine to understand what you're trying to speak or what you're trying to say. But it gets it in line mostly. And it all depends on how you are speaking to it, that is how you're prompting.

Interviewer (7:17)

So you talk about speaking to it. Let's define it. What are you using?

Faraz (7:22)

I'm using multiple tools, Sam. So I started with Midjourney, and nowadays we are literally producing almost like cinematic sequences. So rather than giving our clients just the renders, we produce the whole experience, the whole cinematic experience. We show the transitions from dusk to night with our lighting design, you know, incorporated in the visuals, and then those transitions can be sequenced with other images of the project itself. And the result is maybe a cinematic clip right from 30 seconds duration to maybe 3 minutes, 4 minutes duration, the whole animation. So I use multiple tools for it. Images are being done in Mid Journey, videos are being done in. There's Cling. There is Midjourney's own engine, which is. Which produces very good results, but it's. It's not as sharp. So mostly my favorite tool for the video sequences that the image to video sequences is cling. It's a tool called cling. And then nowadays Google's VO3 is becoming popular, so we are trying that as well in our workflows. So it's a set of tools. And it's not that I've got just these two, three that I have to use because the landscape is evolving at such a rapid pace that every day they come up with something new. Just yesterday they introduced this nano Banana thing, which Google has incorporated in their new Gemini 2.5 update. I haven't tried it yet, Sam, but I'm seeing the results in social media the same way I saw midjourney back in 2022. Since yesterday, I'm seeing these Nano Banana results. So it's basically the specialization of it is that it can edit the existing images if you want to change the camera angles if you want to show the plan, if you've got the elevation, you can. So it's doing those kind of things. And so that's. The tools are evolving at such a rapid pace. Maybe I would be using it in my next project. And now I'm bringing music to it as well. So I use a tool called Suno. So there you can produce your soundtracks. If you want to add vocals, you can do music as well. Proper vocals and proper tracks. But yes, of course, when it comes to work, it's mostly generating soundtracks suited to the project narratives. So, yeah, so it's a multitude and the tools are endless. Basically.

Interviewer (10:02)

What happens when you open midjourney?

Sam (10:04)

Where do you start?

Interviewer (10:05)

Walk me through. Kind of just like a typical. This is how you build the base.

Faraz (10:10)

Okay. All right. So giving you an example of midjourney, when you open the Mid Journey interface, there are two ways. There is Midjourney webpage, and then there is the Discord server. It acts through Discord Server as well. So obviously Midjourney webpage is the more convenient one. When you go to your Create section, then you are just simply, you just need to define what you want to see. That's it. So in terms of the project, in terms of the lighting that you want to see, in terms of the architectural context. So Mid Journey, you cannot do the actual project renders in Mid Journey, of course, but what you can do is you can produce the mood images. So essentially that's what we are doing, whether it's the metaphorical mood images, whether it's the actual project mood images. So you are using it as your personalized Pinterest of sorts, customized Pinterest. So rather than going to Pinterest and finding the mood images there, if you want to customize your own application, lighting application in a way that suits the project, you can define it there and it will churn out the result to you that you can insert into your project concept, whatever report you are making, you can insert it to that, you can tie it to the actual project. So that's in a very simple way, midjourney for you. You can do a lot of other things as well. You can get in the image references. So what you can do is, for example, you have a project render that already exists with the architects. You pull it in midjourney, and then based on that, you try to prompt some other things, some other details. So it will capture the details from your project render, and whatever it produces, it will still tie it to your project render. So your mood image, the new mood image, that gets produced. It will closely match your project intent. So those kind of things. And then there are tips and tricks.

Interviewer (12:22)

Do you have to ask it to reference? I mean, if you give midjourney an image. Sorry, I really haven't used it at all. If you give it an image, do you have to say, please reference this image?

Faraz (12:33)

No, no, no. You just. There is an option just to kind of insert that image with the prompt, similar to what you have now in ChatGPT as well. You can insert any image in ChatGPT and ask anything. So very much similar. So you just insert that image in the prompt section and then whatever prompt you are giving you can assign now here it becomes a little bit technical. So the default value is basically when you don't assign any particular parameter of how much you want that image to be referenced. So there is a set parameter, I think it kind of 50%. It references your image 50%, it produces a new output. But then you can play around with that parameter as well. You may want it to reference your Image maybe by 100% or maybe just by 10% so you can tweak with those values.

Interviewer (13:34)

That makes sense.

Faraz (13:35)

Yeah, it becomes technical, but it sounds technical, but it's not really that technical. Once you start dabbling with it, once you start playing with it, it becomes very easy and very smooth. And in a few runs you would know that, how it acts.

Interviewer (13:51)

And when you're watching

Sam (13:55)

you put a

Interviewer (13:56)

reference image in, you give it a prompt. Is it easy enough to call out and identify like, I don't like this part of it, Change it to that.

Faraz (14:09)

Absolutely. Okay, it is.

Interviewer (14:11)

And it responds quite well to that.

Faraz (14:12)

Yes.

Interviewer (14:14)

Once you, once you have these mood images, what do you do with them? Okay, so I feel like the fun is just starting.

Faraz (14:20)

Yeah, of course, of course, absolutely. So, yes, you have produced these smooth images and you know that these mood images that you have done, they are somehow tying to the project. So you select a few of them and then start tying those, those images to your actual project in a way that, you know, similar to what you do with the project reference images. If you are doing a normal, conventional way of report, lighting, report, lighting concept, report. So you get images from various projects and, you know, from around the world, which have the similar kind of topology to your new project. So in a similar way you can get those images, you can, you know, reference these images to particular sections of the project or you may take it further, you can build a whole story around it. You can further animate those images to make a video sequence. Of sorts, if you want to, you know, present it in a more effective way. If you think that the client is. If the clients would like to see this kind of a sequence tied to their own vision, you can build a whole, you know, whole video sequence of it. And then the video sequence again gets finally tied to the final output of the image that you are actually trying to achieve, the lighting image that you are trying to achieve. So it's a step by step process. First you produce the basic image, the basic reference image. Then you animate that image, and then somehow that animation ties to the actual render. And that actual render is basically the lighting render, not the daytime image that you got from the architects or from the interior designers. It's the actual lighting render. So you can actually have the whole sequence of it. And you can present to the client. When they say it, when they see it in that kind of a way, it's sort of, you know, it makes a greater impact because they can see that. How you started from a basic story of a metaphor, you transitioned through it, made it practical. And then once it comes to the practical side, you are seeing that, how the lighting is transitioning. You can see that, okay, the daytime is getting shifted into the nighttime. And those are the lighting elements, those are the lighting elements that are getting dimmed. Those are the warm elements, and that's the crown, you know, just, you know, shining brightly. And they connect in a better way. Rather than seeing a static image, if you give them a sequence of things, I mean, it's sort of cinematic sequence, they connect in a better way. They understand in a better way. And now it is possible to, you know, to dictate each and every element of lighting, what you want to do, what you want, in maybe example, as an RGBW element, what you want as a warm white element. So you can play around. You can play around with the images, you can play around with the animation sequences. And it could be just a subtle animation of dusk to nighttime and the basic lights getting turned on, or it could be a flashy animation of transition from day to night with RGB coming on and then pulsating lights here, lasers here, things like that. So, you know, it's funky, it's incredible. I would say.

Interviewer (17:58)

Well, do you think AI understands the nuances of light when it comes to, like, beam spread and field angle and contrast ratios and all these other very fundamentally key elements?

Faraz (18:12)

It's getting better at it. It's already understanding quite a lot. When it started, there was nothing. When I started experimenting in 2022, there was nothing at all. I mean, it just understood very basics that, okay, this is a night shot, that's it. But now it understands those nuances very well. It understands the color temperature differences. It understands what's the difference between an RGBW scenario and a warm white or a cool white static cool white scenario.

Interviewer (18:43)

It can be 4000k.

Faraz (18:46)

You can. And it, and mostly it does, mostly it does it correctly. But. But then if it does not work, then to 4000K you could add certain suffixes or prefixes such as 4000K, afternoon, daylight, or if you want to make it cooler, you could say, you could say that 5000K clinical light. So those kind of things. And that's, that's when. That's basically just a test of your prompting skills. So if you just put in 50,500k, it might not give you exact 5500k, but if you put in clinical, yeah, that's when it gets it.

Sam (19:25)

Hey, real quick, do you find this stuff really interesting like I do? I've put together a cheat sheet, something you can download that has this episode summarized in it and is going to give you a basis for how to get started. Do me a favor and click the link in the show notes or just go to lottie.com forward/show notes. Let's pick back up. I asked Faraz, how well does AI know lighting? Here's what he said.

Faraz (19:54)

Tell it to be a 10 degree beam spread. It might give you a 10 degree beam spread. I don't think that the day is far. It's not that far for sure because when I started it hardly knew lighting. Now it knows lighting. It knows nuances, it knows techniques, what's grazing, what's flood lighting, what's. You know what, Saxon. So it knows all of that.

Interviewer (20:18)

A lot of what you're talking about right now is generative AI.

Faraz (20:23)

Correct.

Interviewer (20:24)

You've mentioned a few tools midjourney being one of the leaders in them. Where do you see the transition from generative AI to AI agents? And I guess before we even dive into it, can you define exactly what an AI agent is?

Faraz (20:42)

I'll tell you. For us, for the lighting team at A7, for myself, AI agent is simply, I would say, the thing that does your boring tasks. Boring but essential tasks. You're talking about lighting schedules, talking about the control schedules, the channel schedules and the whole luminaire data sheets in a particular format. We are still in a testing phase, but we started training the agents, sort of. We customize the agents first. We just told it, told one of the AIs to a purpose. Basically, we defined a purpose that what you need to do is you need to, you would be fed with the luminaire data sheets, the manufacturer data sheets, and from those manufacturer data sheets you have to pull in the information, pull out the information, in fact, to extract all the information to fit in the AE7 luminaire schedule and the lighting schedules. So we started testing that and in the beginning it was very frustrating. But then, as you know, the AI kept getting advanced. Now everything, every schedule is automated by the agents. All you need to do is upload the PDFs and just ask it to do its thing. It extracts the images, extracts all the relevant information, puts it in the right places in the format. Whether it's IP rating, whether it's about the drivers, whether it's about the housing, anything. All the technical stuff it automatically extracts from the data sheet, puts in and then churns it out for you. So there, what you're doing is that you are literally cutting down on that time of pulling in the data sheets, getting in the images, formatting the fonts and text and all those kind of things which required lots of hours. We have cut down upon almost like 70% of those scheduling and, you know, the crazy task hours, the boring task hours.

Interviewer (22:51)

So how do you build this agent?

Faraz (22:53)

You can build an agent by the use of any of these large language models, whether it's ChatGPT, whether it's Google Gemini, whether it's, what's its name, Grok. So they all have this inbuilt agent building modes. So you select that mode, you start giving it instructions, you start feeding in instructions, you start telling it simply, in very, very conversational language that what it needs to do and it starts replicating it, it starts learning. And if it does any error, you, you know, speak to it again, that no, this is not how it is supposed to be. You are supposed to take this value from that particular column, not from this particular column. So next time take care of that. And it very obediently, it agrees. And then that's how you. And then you can finalize it and then use the same agent to do your repetitive things again and again and again with every project, with every new schedule that you have to do, with every new, you know, the whole luminate data sheets thing that you have to do, you can just go to the same agent and it will churn out the data for you.

Sam (24:09)

Do you trust it?

Faraz (24:11)

See, good question. I trust it. Yes, it does make mistakes. I do trust it. But then again, the most important thing is that we need to have the manual checks. You just cannot blindly trust it. I do trust it, but not blindly. So we do have the manual checks in place. So once it kind of gives you out everything, you still do the manual. Just like sort of a proofreading a document. You cross check each and everything and most of the times it does not do any mistakes. And the rarest of times it does, it's mainly something minor which you just again, when you say that this is not supposed to be like this, then it kind of, you know, rectifies it. So yeah, I do trust it. Not blindly. And again, the human element has to be there for final prospects, for final proofreading. But again, even if you are proofreading it, you are still not spending as much time as you would do in building that particular document, in making that particular document. You are just doing a sanity check, a final sanity check. And then, yeah, there you go.

Sam (25:29)

Before we keep going, I want to spotlight someone who shares the same obsession with the design detail that this podcast is built on. Kelvix doesn't just move fast, they move with purpose, helping the design world turn ambitious ideas into seamless, buildable realities. The Kelvix team works at the pace of your project, delivering tailored solutions backed by the people who know how quality, timing and precision elevate the final outcome. In every spec and every collaboration, Kelvix proves lighting is more than illumination. Lighting protects design intent, sharpens the details that matter, and brings spaces to life

Interviewer (26:06)

with clarity and ease.

Sam (26:07)

See what's coming next. From breakthrough products to standout projects@kelvix.com or on their social channels. I also want to take a moment to share another company that I respect and enjoy learning from at LED Flex. Linear Light is designed to do more than illuminate. It's engineered to shape how spaces are experienced. Rooted in British manufacturing, LED Flex works alongside lighting designers, interior designers and architects worldwide, supporting projects where detail, precision and execution matter. They're not just a manufacturer, they're a technical partner providing bespoke solutions, rapid prototyping, commissioning and long term support from early concept through final handover. If light is integral to your architecture, choose a partner who stays with the project and learn more@leadflex group.com there's also one more company that I want to introduce you to because they consistently show up where it matters. I know and you know every designer knows the success of a space depends on actually what shows up on site. And that's where Diode LED comes In lighting performance isn't an add on. It's the foundation of every successful linear lighting project. Designers need solutions that carry ideas smoothly from specification through installation. And Diode LED is built to support that reality. With industry leading inventory, reliable delivery and responsive support, Diode LED keeps pace with real project timelines, not theoretical ones, real solutions for real projects, no matter the scale. Learn more about what's happening in Reno, Nevada at Diodeeled.com Pretty neat, huh? All right, let's get back to the conversation.

Faraz (27:53)

Believe me, I mean we have built agents in three working days. We got the luminaire agent up and running, the luminaire schedule agent and that's it. I mean there are. Because there are many trials and errors that happen and you know, it might not get things right at straight at first, so you have to maybe speak in a different way to it. So it takes a little bit of time in just building it, but once it's up and running, once it's there, it's very quick and it's getting better with time. It's constantly evolving itself. It's constantly learning from our own data, basically. So it's learning fast and it's learning well. So nowadays if I have to build something new, I don't spend as much time as I was doing back in 2024. Yeah, that's it. Just about maybe six months ago. It's weird, it's really, you know, the exponential curve is, you know, it's crazy.

Interviewer (28:56)

What's the one thing that you don't want AI to be able to replace that you know it's going to or already has?

Faraz (29:06)

The one thing that AI should not replace, right? The human intuition, the human sensitivity. You cannot and you shouldn't. The human factor is. It has to be there. You cannot see. AI is like I said in some of my earlier discussions as well, I have said this, that AI is basically not a replacement. It's just a tool. It helps you do things in a speedier way. It helps you in doing the iterations fast in the concept developments and the story development. But it's not a replacement, it's just a pencil. It's just a copilot. And if done right, it's amazing. If done wrong, it can go bad. It does go bad. It hallucinates by itself sometimes. And of course then there is this thing of. Because these are trained data sets, so they tend to be biased. So you have to, in order to check all those things, you have to have that human element at the forefront. That human has to drive that and AI is basically, it's just a power tool that's happening side by side.

Interviewer (30:28)

I've never thought about this, but does AI know what AI is?

Faraz (30:33)

That's a great question. I am even I cannot answer this because I think it's a question for the machine learning experts and I believe even they don't have the solid answer. Because what I know is that it works like a human brain, it adapts, it learns, and pretty much works like a human brain. And if you see at the back end, what goes on, you know, at the back ends, it's pretty much quite similar to what happens in our brains. The thought process, it's quite similar, maybe amplified million zillion times, but the process is similar. The, you know, the whole, you know, the sequence of it is much similar. So I think maybe it knows because if you ask and if there is something that it cannot do, so it does say that, okay, I'm just a machine learning model, I cannot do that, I'm not able to do that yet, or something like that. Maybe it knows what's the good and

Interviewer (31:42)

the bad of AI entering into the workflow that you're implementing right now.

Faraz (31:49)

So the good is of course, the things that I explained a little bit earlier that it, you know, very fast, you can create a whole story, a sequence of events which, you know, helps you, helps your clients understand the project design intent in much, in much better way. So you have these sequences, these images which can be very easily understood by clients. It helps in fast approval processes, it helps in reduced, you know, the reduced times on our ends. We can, we can do more concept exploration in much lesser time and much faster, which is good for business, of course, and everyone likes it, our bosses like it. So they're glad that, that we can, you know, sort of contribute in that way. That, so those are the good things and of course the other things. When it comes to the agents, you can very quickly do the, the boring tasks as well, very quickly. Effectively. You have to check. But then again, even if you are checking, even if you're cross checking each and everything, it doesn't take as much time as you would do that manually. And the bad is that if it deviates, you need to know when to stop. Because the data sets are vast, they might have a bias, they might be bringing in some characteristics of an artist or an architect which is very much identifiable. So you have to be mindful of those. And that's when the human again comes in. You have to keep that in check. That as soon as you Feel that the prompt is going a bit haywire. You have to stop it, you have to take a step back and maybe start again. So yeah, that's the bad side of it. You have to stay ethical, you have to stay clear of any copyright issues. So you have to stick to the. There are hundreds of tools out there, There are just a few which are, you know, I would say you can count on. So you have to stick to license tools. You have to do your outputs in a way that it does not infringe on any others copyright, it does not copy anything from any other, and it's very much capable of it. So what we do is how we overcome this aspect is that before starting on the AI, we do one page cheat sheet. Cheat sheet of sorts. There we define the guardrails. That AI is not supposed to do the no go zones, for example. You cannot copy any architectural styles of any famous architects. You cannot just, you can think from your side, but you cannot copy any building. You are not supposed to. So when it comes to lighting, you're not supposed to apply it, you're not supposed to show up light, you are supposed to remain low key with warm color temperatures. And those kind of things we can define on a cheat sheet. Sort of a, sort of initial narrative for the project, for the images that we want to see. And clearly the no go zones are defined there. And then when you do that, it's very much controlled. You can control it to a great extent.

Sam (35:24)

Do you tell it, do you tell it, this is the constraints, this is the cheat sheet? Or just, just. Or do you just say, hey, take this all into account?

Faraz (35:31)

Yes, that's what we do. I mean, we're not saying that it's a cheat sheet, but for example, we don't want it to copy any certain style. We have to specify, we have to mention that. And when it comes to the lighting techniques, you can easily say that avoid up lights, be gentle, grazers are okay, but just maybe just to a height of about 1 meter, not above that. And those kind of things you can tell it and then it's very much controlled.

Interviewer (36:05)

Is there a cultural risk or, you know, a disappearance of culture that converges because of this?

Faraz (36:12)

It could, it very much could. But then that's where, that's why the human element is important. That's where you have to keep it in check. You cannot let that happen. So you have to always see that your outputs are not out of cultural context. They still fit in the cultural context. If they do that, that image will not ship. So you again, those are the things, those are the human traits that you have to be on top of it. It's very like you said that it's. The world is a data set. It has data from everywhere. So it happens that it could hallucinate. You're telling it to emulate something, you know, maybe. Maybe build something here in Dubai, and it kind of, you know, gets in the architectural styles of Japan or China. It is possible sometimes they go haywire. And that's when you know and, and that's when your own knowledge helps to guide it correctly. You know that, okay, when you, when you see that image from China, you would know immediately that, okay, this is. This does not feel right. This cannot ship, reject, restructure the prompt or just redo the prompt. Most probably in the second or third attempt, it might get it with the same prompt.

Interviewer (37:30)

What's the one thing AI can't do today that you really want it to?

Faraz (37:36)

The production side of it. See, aesthetics is already pretty much quite. It's there to quite an extent. You can do the renders, you can change the renders if you want to. You can edit just like Photoshop. You can edit anything that you want to. You can do these multiple, like, cinematic features if you want to. But on the production side, I think right now it's what, what we need is an integration into the production models like revit, like dialects, where you do the actual photometric. Photometric calculations. So I'll give you an example that for dialects, we are pulling in all these luminaires when you are doing, when we are doing the luminaire calculate the project calculations, lighting calculations. So with integration of some sort of an AI, there might be a possibility that, you know, you just feed the, the whole project data to the two dialects and then it kind of automates everything within itself. Of course, you still will remain in control, but that, that first iteration, which will take you time to select the luminaire from this particular manufacturer, from that particular manufacturer, whether it's a downlight, whether it's an uplight, or, you know, CCT 3000K, 4000K.

Interviewer (38:59)

So opening up your Revit model or your PDF and saying, go build the baseline photometric model.

Faraz (39:08)

Absolutely.

Interviewer (39:08)

And run an initial pass and it goes and finds the IES files, downloads everything, catalogs it, names it, check, references it off your luminaire schedule, and then

Faraz (39:17)

poof, there you go.

Interviewer (39:19)

Human element, you know, sniff, test.

Faraz (39:22)

Exactly, exactly. And the same thing in the production drawings as well. And I think some of the interfaces are already out, but we haven't tried it yet, but I think it's just a matter of time. So once that happens. See, the idea is that the thing is that when it comes to AI, the first thing that comes to your mind is, okay, not a replacement, but an assistant. So it is already assisting you in the early project side of things, from concept to sd. Maybe it's helping you a lot in doing a lot of things when it comes to the actual production, most probably, I mean, it's just a matter of maybe six months to a year, those things will come in. And yeah, I don't think that it's going to take more than that. And people are still. I mean, I think they are already testing out a couple of things with Revit. And, you know, there are Autodesk is testing out some interfaces. So once it's. Yeah, I think it's just a matter of time. So that automation, which is already happening in the early project stages, that automation, which will also be happening in the later project stages when you are actually doing the whole production work, laying out all the luminaires. So most probably the lives of the BIM modelers would be getting easier right now. It cannot do that, but I think it's just a matter of time.

Sam (40:50)

I want to take a moment to recognize a team that understands what every designer already knows. Light isn't a fixture, it's the soul of a space. For almost a century, Targeti has shaped how people experience architecture using precision optics, performance, and that unmistakable sense of Italian design to elevate every moment a space offers. Because light isn't just seen, it's felt. It drives emotion, guides the eye and transforms intention into atmosphere. For Targeti, light isn't simply the medium. It's the identity, it's the craft, the legacy and the commitment behind every solution they create. Experience it for yourself. Visit Targetiusa.com now, let's jump back into this conversation.

Interviewer (41:39)

I think the last thing I have to mention is we're recording this conversation at the end of August in 2025, because I know that it won't be published tomorrow.

Faraz (41:50)

Right.

Interviewer (41:50)

And it's very likely that between now and then there will be measurable differences.

Faraz (41:56)

Measurable.

Interviewer (41:58)

And that's. That's just something that I guess is cool but crazy.

Faraz (42:03)

And. Yes.

Interviewer (42:06)

Would you say that, you know, you're a. A duck in a pond or a fish in the sea at. When you look at your company and how many people are. Are dancing with AI right now?

Faraz (42:19)

Oh, everyone. Everyone is adapting to it now. So when it comes to my team, everybody is already adapted to it. All the designers are using it now. Even the BIM modelers and the Revit technicians, they have started playing with ChatGPT and Midjourney and even if they have any technical revit issue. So before going to our own Revit team, they actually Dabble in with ChatGPT and get the answers themselves. So that helps. So they're getting savvy, they're getting AI savvy and it's all through the company. So our company itself, they have this, you know, I mean, it's almost sort of an, you could say that it's a mandate now to, you know, follow the AI workflows. Everybody is adapting to it and it suits the workflows very well. It suits the architectural workflows, it suits our interior designers, it suits our landscape designers. So everybody is much into it and we do a lot of cross disciplinary exchange as well. There are talks, there are things that they share that we do not know. We share something that they do not know. So we keep having that dialogue open between all the disciplines and our principals and higher management, they have been promoting this since a long time. So yeah, I think it's. Everybody has to adapt to it and I think no one should be left behind. The only guys, I mean, you know, people who say that AI will replace the only people who will get replaced would be, of course, the ones who would not know what AI does and what it is capable of doing and the versatility of it and the power tool that it is. That's, that's the thing.

Interviewer (44:22)

Don't be afraid of it. Embrace it.

Faraz (44:23)

Embrace it. That's it.

Interviewer (44:25)

Faraz, thank you so much.

Faraz (44:27)

Thank you, Sam.

Interviewer (44:27)

I appreciate it. I'm excited to see a little bit more of what you're working on.

Faraz (44:33)

Absolutely.

Interviewer (44:34)

Follow along as you continue to share, you know, your advice and experience with the community. But most importantly, see what everybody in the design world can do. As you pointed out, we embrace this and grow with it.

Faraz (44:46)

The tool itself is rocking, Sam. It's, you know, it sort of, you know, empowers the creativity, I would say. And it makes you, you know, shine better if used properly, with constraints, with guardrails, with the responsibility that makes sense.