One of the things I appreciated most about talking with Cecilia Uhr was how much her story reflects the reality of being a designer founder today. She started out doing packaging design in art school, exploring her curiosity about how people experienced a product, not just how it looked. That instinct has carried through everything she’s done since: agency work, AR/VR at Oculus, and now co-founding Bezi. Bezi’s current product—a Unity dev assistant powered by real-time AI context—wasn’t where they started.
In my recent conversation with Cecilia, she walked me through their pivot and the tough decisions that come with trying to build something useful in an emerging space. We also talked about the shift from agency to product work, learning to prioritize as a founder, and why customization is the next big frontier for AI products. There’s a lot in here for anyone figuring out how to build under constraints, trust their taste, and adapt when the work demands it.
What we cover in our conversation:
The strengths of an agency background
Cecilia’s lessons from transitioning into tech
Knowing when it’s the right time to launch your own company
Founder skills designers already have—and the skills they need to learn
How to weather hard startup decisions
Cecilia’s biggest learnings in building Bezi
Why brand and storytelling are more important than ever for AI companies
How did you first get into product design?
I went to an arts high school where I majored in visual arts. As much as I loved drawing and painting, I realized it's really hard to make a living off being an artist, so I pivoted to design.
At that time, the idea of “product design” was just beginning to emerge. I didn't really know what it meant, but I knew that I liked designing experiences. In one of my college packaging design classes, I realized I was less concerned about the aesthetics of the package and more interested in the experience of someone opening the package. What is it like seeing the package on the shelves, picking it up, and interacting with it?
When you first started working, you chose the agency route rather than immediately getting into tech—what appealed to you about that?
I like doing a lot of different things. Going the agency route, I wanted to get a breadth of experience rather than just working on one feature for several months. I wanted to be able to work on the brand, the website, and the product all in one place.
Looking back, that was the right choice—at least for me. There's a certain level of grit that comes with working at an agency. There’s a healthy competition with the other designers you work with, and you’re working a ton of hours. You get to try a bunch of different things, and you build a good portfolio because you're pumping out so many concepts.
What made you switch to tech?
I wanted to be closer to the work. At agencies, you're tasked with coming up with a concept and after you hand it off to the client, you don't see it or touch it again. Sometimes the idea gets mutated along the way—sometimes for a good reason—which can be disappointing.
I wanted to be part of the whole process: collaborating with engineers, getting user feedback, and iterating on the product. I wanted to see things through to the end, not just hand them off.
So I started looking into ways to get into the tech industry. I was lucky—I didn't have any AR/VR experience at the time, but I ended up at Oculus. I started off as a visual designer, and then transitioned over to a product design role.
Know a designer founder we should feature next?
We're always looking for inspiring designer founders who are building something meaningful. If you know someone who's made the leap from designer to founder—whether they're just starting out or scaling up—we'd love to hear about them.
What were some of the things you learned going from the agency world to a software startup?
Agencies are usually optimizing for awards and recognition because that's how they get clients. That's the mode that I was operating in when I joined Oculus. I was so excited by the possibilities of the product. And then of course one of the big mistakes I made was not taking technical constraints into account. I came back with all these ambitious concepts for new user onboarding, showed them to the engineers, and they were like, “yeah, that’s not going to happen.” It just wasn’t technically feasible.
That was a hard lesson to learn. I remember at the time feeling very small, asking myself: why didn't I think of this? But when you're working at an agency, you don't think about that at all. You're just like, how do I make the most eye-popping, awesome experience possible? And that's just not how it translates to the software world.
When did the idea for Bezi start to take shape?
Bezi stemmed from personal frustrations between me and my later co-founders, Julian Park and Denys Bastov, and needing to find a way to collaborate with them while working at Oculus. As a designer at the time, there were a few options for designing AR/VR experiences. You could work in Figma, a tool you're familiar with, and try to mock up 3D-esque designs—but these wouldn’t truly be able to replicate a sense of space, distance, or speed of movement. You could try to mock it up in Blender, but you’d end up having the same issues—seeing something 3D in a 2D frame. Or you could learn a hardcore game engine like Unity to actually be able to design in 3D. The problem with that is that it’s not something already within the designer’s skillset—or even job description—so many designers are allergic to learning it.
So, we built a web-based, collaborative editor that made it easy to design and prototype experiences in 3D. We initially did it on the side on nights and weekends, excited by the idea and having fun exploring where it could go.
When did you know it was time to start your own company?
It was a building feeling rather than an inflection point. It's not like we were trying to get users or anything at the time. We were just focused on trying to solve our own problems. We eventually reached a point where we realized we were more excited about this than what we were working on in our day jobs. We were also frustrated with the layers of approvals and how slow things were moving. We decided to go all-in on building this new company.
How did you decide on the roles each of you took when launching Bezi?
It came pretty naturally to us and I think that’s because we had worked together before. We understood each other's strengths, what each of us was interested in and what we were not interested in. For me, coming from a design background, it made sense to take the lead on design and product.
You ended up shifting Bezi’s focus away from being a collaborative 3D design tool to being AI-powered dev tool. Walk me through that process.
For the record, I still believe that the first product we were building will become useful at scale at some point. The hard thing about that is that AR/VR is so dependent on hardware and technology. So if the hardware isn't there yet, if these AR glasses can’t function like we need them to, then mass adoption doesn't happen and we can’t be successful. It’s entirely out of our control.
We asked ourselves: does it make sense to keep working on this, or should we tackle this from a different direction?
Honestly, it was a really hard decision—probably the hardest decision that we as a company have had to make. We decided to move away from the 3D editor that we had been building, while still focusing on our mission of making it easier for people to design experiences and games. That's the new product that we're building right now, and we're leveraging some new technologies I’m really excited about.
What helped you make that transition?
The startup journey is all about sequencing and prioritizations, and sometimes you just need to make hard decisions. We had to do a lot of introspection, and decided to have a small 'tiger team' that prototypes the new product in the beginning. The rest of the team was business as usual during this phase, and we shifted the whole team around to work on this new product as we saw promise.
As a team, we had to learn a lot. None of us had experience with AI or LLMs. We had to look at a lot of existing products out in the market, understand how things work, and just try. It was a lot of prototyping, and a lot of trial and error. My hope is that what we had built before can be integrated into this new product, which could make it really powerful.
What were some of your existing skills that have translated well to being a founder?
I worked on brand and marketing materials when I was at agencies, and that helped a lot as we built Bezi. I made the logo, designed and built the website, all of that stuff. If I didn't have that experience, I would be struggling a lot more now. I believe that brand is becoming increasingly important in this new world of AI products. How you show up, how you communicate what you believe in, and the storytelling behind these values is how you differentiate your company in a sea of sameness.
At Meta, I learned a lot about how to work with cross-functional partners like engineers and product people—understanding what their roles are, why it's important to the company, and how that all ladders up to business goals. I was able to use that knowledge to understand what we’re missing, so I know what to hire for.
What skills have you learned as a founder?
The weight of decision making, and prioritization. Those go hand in hand because you need to make decisions on what you're going to prioritize and that could sometimes determine the trajectory of your company.
One of the things that designers actually struggle with—and I'm speaking from firsthand experience—is that prioritization is really hard. As designers, we're trained to craft and design the perfect experience. We call it the happy path for a reason! You want to make sure that people have a really good experience and they're able to achieve their goals in your product.
The truth is that startups are always operating under conditions of extreme constraints. So it's very unrealistic for designers to get everything they want in a feature or a product. In those cases, you need to be able to differentiate between deal-breakers and nice-to-haves. It's challenging to let go of some of these things that you know will make a delightful experience, but that aren’t integral to the product or to the feature, but that’s a muscle designers need to build to be able to ship fast and iterate.
Do you have guiding principles that help you make the right decisions around what to prioritize at Bezi?
When it comes to high-priority features, we always need to figure out: what is the main goal that the user is trying to achieve, and will not shipping this specific thing deter the user from getting to that point? Or are there ways that they can get around that will still help them achieve their goals? So if it does allow them to do that, maybe it's not a P0, maybe it's a P1 or a P2.
An example of this is our onboarding experience. This might be controversial because the onboarding experience is your product’s first impression, but the product we're building now relies less on an onboarding experience because it's already so intuitive. It's a chat interface that most people are familiar with. Who doesn’t know how to message someone, right? Because Bezi’s core experience is so intuitive, we decided to cut the onboarding experience.
Once we shipped this first version of our product, we got feedback that some people needed instructions on how to connect their Unity project. So we built it in a day and shipped it. This new small improvement cut about 90% of the negative feedback around onboarding.
That’s a good example of where you need to identify a quick improvement you can ship to help 90% of people get what they need to use your product. And then over time, when you have more resources, you can refine it.
Even though you might have a smaller audience than Oculus or Meta, it can feel like product decisions are even more important. Why is that?
At a tech company of that scale, even if you make the wrong decision, it's usually not existential. Google ships hundreds of products every year, and they scrap 90% of it. They have the resources to be able to do that. For a startup, if you make a wrong decision, it could be existential. That's something I grapple with every day.
You have to be sure that the decision you're making is the best decision at the time, and that you understand what the risks and the consequences might be. I can't remember who said this, but it resonated with me a lot: some decisions are one-way doors, and some are two-way doors.
You also need to learn how to decipher what is actually important. It’s a key skill for a founder to have because if you don't, everything is urgent. Everything is a crisis and your team will feel that.
What has been one of your biggest lessons in building Bezi?
When you’re building a 3D design tool, you’re essentially competing with products like Maya or Autodesk—products that have been around for decades. Unless you reach somewhat of a critical feature parity, your platform is useless to people because they can do it elsewhere much faster. Design tools are just generally incredibly difficult to build, and I have a lot of respect for anyone who's doing it.
One of the mistakes that I made early with Bezi was trying to get to that feature set without a highly prioritized strategy. I wish that we had started with a more solid work-back plan from where we want to be, but it's a lot harder in practice when you're getting feature requests left and right.
How do you think about the role of craft in a startup?
My designer Daniel Marqusee and I had a debate about this recently where we were talking about the word “craft.” I do think it's a very overloaded word that needs to be defined specifically or else it can be weaponized a little bit. Often, people will say, “We need a higher level of craft on this.” And then you’re like, “What do you mean by that?” And they can’t clearly define it. You need to be intentional about what kind of experience and feeling you want to evoke in your users or your customers—and that can look a lot of different ways.
What role does AI play in your brand and storytelling?
Nowadays everyone's kind of working with the same materials. Everyone has access to the latest models, and we can kind of bend and mold them how we want them. AI has leveled the playing field.
What we’re focused on now is how do we test the limits of the model, and how do we make sure people can achieve their goals going through it? Brand and storytelling play a bigger role than ever. In a world where everyone's building a chatbot, how do you stand out? You need to invest in how you tell your story.
We're going through a rebrand where we're trying to imagine what Bezi feels like with this new product that we're building. A lot of the actual craft is going to come through product capability and personality, through the chat interface. It's been really interesting because we're talking about how do we want Bezi to feel—and it's no longer just about aesthetics.
What are some of the biggest challenges of designing with AI?
I think customization is going to be the next frontier of AI products in general. Because you can have a very distinct personality for your product, not everyone's going to be happy with it. Remember when OpenAI shipped this new update to the personality where it was very proactive? People hated it.
I’m thinking a lot about how we allow people to customize their experience. To set not only the personality, but also the rules and instructions for how they like to be interacted with. Designing that is really interesting because it’s something that hasn’t really been possible before.
There’s going to be a tension between the brand personality and what we allow people to customize. You need to have a point of view that people can tie to your brand and your product—but you also need to give people enough leeway to be able to customize their experience and interactions with your AI.
How do we bridge that gap? I don't know what the right answer is right now. It's something that we're trying to figure out. I'm also seeing every company in this industry try to figure that out too. That's why a lot of companies are going this neutral route, because then it can be filled with whatever meaning the user wants it to have. But that also has its pitfalls. You can start to blend in.
What’s inspiring you these days?
I’m having fun trying all the cool products that people are putting out. Some of the most interesting ones currently are node-based creative tools (ComfyUI, Flora, and Weavy), and seeing how chat interfaces are evolving (ChatGPT, Claude, Gemini). That's what I love to do—trying to figure out why the team made certain decisions and what the product is capable of doing.
With the amount of new products and models coming out every week now, it's absolutely insane. But I think that’s what makes it a really exciting time to be a design founder. The playing field has leveled out. You can try a lot of different things to see what works. We’re at an inflection point, and I love that everyone is learning together.
Books that have guided Cecilia
Building a Story Brand by Donald Miller
Traction: How any Startup can Achieve Explosive Customer Growth by Gabriel Weinberg and Justin Mares
Originals: How Non-Conformists Move the World by Adam Grant
This is Service Design Doing by Marc Stickdorn
