How AI will influence creative tools: A conversation with Replit’s David Hoang
Replit Vice President of Marketing and Design David Hoang talks about how AI is reshaping the future of product design and development, and the role it’s playing in the company’s team and products.
When Apple debuted the iPhone in 2007, David Hoang was still early in his career, but he saw a window of opportunity in that transformational device. If the iPhone succeeded, the designers of its new interfaces would shape a new era of human-computer interaction. He doubled down on learning how to build for mobile apps, and he hasn’t looked back. His career spans over two decades of product development experience, from starting his own studio, to teaching hundreds of students UX Design at General Assembly, to working as the Director of Product Design at Webflow. His current role as the Vice President of Marketing and Design at Replit, a software creation platform, puts him at the forefront of the way we make products, including how product builders and users will start to use AI.
We invited David to Figma’s New York City office to discuss a range of topics at the intersection of AI, design, and product development. The following is an edited version of that conversation with Mihika Kapoor, a Product Manager at Figma.
How do you envision AI changing the way we design and build?
When the mobile boom started, everyone who was an expert became a beginner again. In a way, the paradigm shift reset the playing field. It was an opportunity to imagine a new interface paradigm—a designer's dream—and it feels the same way with AI. We are now moving to a world where we do not fully design the interface presentation anymore for people who experience it, because we are going multimodal, multi-form, factor, what I like to call dynamic interfaces. To be able to control where things are going, you actually have to let go of control of that instinct. It’s a little scary but also exciting at the same time. AI, spatial computing with Apple Vision Pro, and many of these other trends are not parallel tracks that steer independent of each other; they are converging.
How do you think AI will impact the relationship between how people design and how people code?
Whether you currently work more in engineering or design, AI offers augmentation for you to do both. Engineering and design are on a course to become one tightly woven discipline. At Replit, we’re focused on building Artificial Developer Intelligence (ADI), not Artificial General Intelligence (AGI). We want to build something that gives people more autonomy and makes them more productive. Collaboration and AI, these are two very foundational pillars in how we think about our ADI strategy.
Replit’s Vice President of AI Michele Catasta wrote a manifesto about artificial developer intelligence: “In time, the Replit ADI will create Agents capable of developing complex software architectures, both by generating code and by orchestrating advanced software tools developed and deployed on Replit.”
What are some examples of how people will be able to work with ADI?
ADI might mean code completion, and how you generate code. It might augment organizational intelligence, and understand the context of how teams work together and how they can better collaborate. I’ve always viewed Replit as the technical co-founder people need, and I’m starting to see a lot of people treat Replit like that, too. While it’s great that there are a lot of people learning to code, we need to figure out how to accelerate from learning to shipping, to building a business or launching your idea, and growing it as fast as possible.
For example, in the non-technical track of a recent hackathon, one of Replit’s Builders, Priyaa Kalyanaraman, used Replit and AI to build a platform that would inject documents with personality through animated GIFs, fun language, and text-to-speech. AI wrote 100% of the code. After winning the hackathon, Priyaa started up Lica and gained funding for her idea.
Do you have a dedicated team working on AI at Replit, or are there multiple teams trying this out?
We have a dedicated, centralized, AI team that does a lot of training for our large language model that we’re hoping will grow through product development research. This team also helps us understand what’s important for end users when they use AI. Even though we have a centralized team, everyone is thinking about how AI will apply to their respective jobs.
How much of the work that you’re doing is experimental versus well-defined in terms of the ambitions and product development goals?
When we’re thinking about APIs and services, I don’t think we can be too experimental. Similarly, if you’re doing a language model, that’s going to take more time. We need to make sure these aspects are done right, done thoughtfully, and that they can scale. There’s room to experiment with the interface, because I don’t believe the interface paradigm has been defined yet. A large percentage of that is really thinking about what the paradigm will be and how people are going to respond to it.
Does AI eliminate the role of the designer?
Between the current job market, and the things AI can automate, being a designer can feel scary right now. However, my fundamental belief is that designers are always going to have a purpose and a role, because there are always things that will need to be designed. I encourage people to consider that design may not be literally what we’re doing today. Design will shift and evolve, we should approach it with an open mind.
Between the current job market, and the things AI can automate, being a designer can feel scary right now. However, my fundamental belief is that designers are always going to have a purpose and a role, because there are always things that will need to be designed.
What role do you think designers will play in this AI revolution?
When technology takes over in terms of automation, it often requires more critical thinking and creativity to think about what’s next. That’s an exciting thing for designers. At Replit, we feel like we have an opportunity to contribute and define what AI will become. I’m sure the Figma team feels similarly. People often say things like, “I need to get into AI, or I’m going to fall behind.” My usual response is, “We are designers, we’re built for this.” Designing an AI is just systems design. You’re doing the right things already, you just need to apply it to a different technological challenge and really explore that.
What is your advice for designers who are excited to jump into the AI space but are not quite sure where to start?
When it came to bringing new AI features to FigJam, the Figma team focused on defining and packaging the elements of a good FigJam board into core principles that could apply to a wide range of user needs.
Make time to play with products and begin to discern between what quality AI use cases are, and which products are using AI as an afterthought. You want to aim to really understand what AI can augment, automate, and replace. It will be even more important for people to contribute where AI can’t. The barrier for AI is not technological advancement, but the ability for people to learn and adopt new behaviors. You can have the most innovative product, but if it’s something individuals and companies don’t adopt, it’s going to be extremely difficult to succeed.
Are you changing anything about how you’re running your design team or how you’re hiring with regards to your ADI strategy?
At Replit, we’ve always leaned towards hiring multidisciplinary designers who understand the technical depth it takes to create software and bring it into the world. Each designer has the output of three to five designers that you’d ordinarily work with. They code, prototype, review pull requests, do unit testing, facilitate research sessions, and run workshops, too. Our product managers and our engineers also have great design taste, and are very design centric.
How do you manage building on top of technology like AI that’s evolving at such a rapid pace?
To stay on top of the changes, the best strategy is to be adaptive and really gain a deep understanding. If you’re online, you may experience a sense of imposter syndrome, because everything feels so overwhelming, and people express strong opinions. In reality, the playing field has been reset, so this is the time to experiment a lot. We don’t know what’s going to stick around yet.
What do you think is special about the Replit community?
There are the people who go through the instructions, step by step. I am not that person, I just open the thing and figure it out. For people like me, we really want to meet others where they are on their learning journeys. We want to build things and ship them. We want to do it tonight, and get it right. This is something Figma does so well. There’s a feeling of being in a cohort; a balance of accountability and fun. It’s gathering a community in a way that encourages shipping and iterating. That’s going to be what continues the motivation to learn.
Read more about building with AI here.
Related articles
