Dive Club 🤿 – Ian Silber: What it's Like Designing at OpenAI
Host: Ridd | Guest: Ian Silber, Head of Product Design at OpenAI
Date: April 8, 2026
Episode Overview
In this episode of Dive Club, host Ridd talks with Ian Silber, Head of Product Design at OpenAI. Together, they go deep on what makes the design environment at OpenAI unique, how designing with AI as a material has transformed the design process, what it means to be a systems thinker in the age of AI, and how the designer’s role is rapidly evolving. The conversation touches on the transition from traditional UX/UI to model-led product design, the unique challenges (and excitement) of building in a landscape that shifts almost daily, and advice for designers looking to thrive in this new terrain.
Ian’s Path to OpenAI
[00:51–05:10]
- Began career at Instagram, notably involved in major projects.
- Joined Artifact (“a sort of news kind of AI thing”) with former colleagues and friends, blending Minecraft-like ease-of-use with Roblox’s ecosystem.
- “I was at the point in my career where I was like, I should just try some crazy stuff... I was kind of unqualified. But I mean, the team... was both really close friends... but also some of the most talented people that I've had the chance to work with.” (Ian, 01:42)
- When GPT-4 emerged, Ian’s team started using AI tools to power their startup, which caught the attention of OpenAI’s Head of Product.
- The entire team—designers and engineers—was recruited as a group to OpenAI.
- “We went from, like, talking about some random specific game mechanic on like a Friday and then like that next Monday... deep in the OpenAI trenches, trying to figure out what we're launching next.” (04:44)
What Makes OpenAI Design Different
[06:53–07:59]
- OpenAI is fundamentally a research lab—a major contrast with consumer-focused companies like Instagram.
- “From day one, it was just such a different place... the rate of progress and what is coming... this place is different.” (06:56)
- Deep mission-driven culture, with emphasis on what’s possible with advancing AI models.
- The challenge and excitement stem from rapid, ongoing change and technological evolution.
Designing Closely with Research
[08:10–09:38]
- Designers are as close to the AI model as to the UI, always “playing with and trying it, seeing where it breaks.”
- Curiosity is more important than technical depth; designers translate research breakthroughs into user-facing features.
- Emphasizes thinking of the model as the product: “A lot of times, like, what can we do this without pixels? Can we do this with tokens?” (08:40)
- Pushes for solutions that use the model directly rather than elaborate UIs.
Design “Beyond the Pixels”: Real-World Examples
[09:38–12:00]
- Rethinking onboarding: Instead of traditional walkthroughs, leverage the model to infer user needs dynamically.
- “We’re really like, stripping back a lot of maybe kind of what you might traditionally do and trying to say, well, actually let's think about how we should give this context to the model...” (10:30)
- Designers focus on the system prompt, model behavior, and context—to change product behavior.
When to Build UI vs. When to Lean on the Model
[12:00–14:51]
- No strict principles yet—often an intuition-based process.
- Case Study: Writing in ChatGPT.
- “We looked at the data of how people use ChatGPT. A huge percentage of people use ChatGPT to help them write... We wanted to kind of lean into more direct manipulation.” (12:55)
- New container-based writing UI allows fine-grained edits directly, blending model behavior with ergonomic interface.
- Aim to build systems where the model itself can dynamically compose UIs suited to user tasks.
Systems Thinking at OpenAI
[14:51–16:49]
- Best systems thinkers think not just about features, but about extending the entire platform.
- Focused on creating powerful “primitives”—reusable abstractions like “skills”—that underpin the system.
- “We want to have a system that not only a human can understand, but a model can reason about.” (15:38)
From Raw Experiments to Shipped Product
[16:49–18:35]
- Prototyping tools (like Codex) allow anyone—designer, engineer, PM—to build and demo ideas quickly.
- Bottoms-up innovation: “There's a lot of pockets like that, there's a lot of bottoms up. Just in general, that's another thing that's, I think, different at OpenAI...” (17:49)
- Key: Empower everyone to build, but edit ruthlessly for systemic coherence.
How Design Practice Has Shifted
[18:35–22:40]
- “Instead of a Figma prototype... it's like very interactive. I can go and just tap around... actually in many cases I can start asking questions or... see what the actual model behavior would be.” (19:37)
- Internal agents empower designers: “Any designer working on something can ask questions, well, how do people actually use this?” (20:10)
- Choosing the right tool for the right fidelity: from sketches to production-ready code.
- “Sometimes you want a whiteboard, sometimes you want a paper sketch, sometimes you want a wireframe... sometimes you want these live prototypes.” (21:33)
The Evolving Skill Set of Designers
[22:40–26:28]
- The industry has swung from hybrid designer-engineers, to specialization, and now is swinging back due to accessible prototyping with AI.
- “It's kind of swinging back where I think both, both can overlap way more. And so there's just a big question around what skill set...” (23:56)
- Core design values—problem-solving, craft, taste—remain constant, but flexibility and adaptation are crucial.
- Tools (Quartz Composer, Origami, Codex) defined each era; now, AI-powered tools enable new creative expression.
Designing Not Just the Shell, but the Content
[26:28–27:09]
- OpenAI sits between pure shell (like Instagram) and total content control: model outputs are part-design, part-unknown.
- “You can play a little bit with how the model should behave... it's somewhere in between... totally different.” (26:57)
Building and Scaling Design Culture at the Speed of AI
[28:23–31:27]
- Organizational shifts: Only now building a design systems team (“Dynamic User Interface Library”) to support scale.
- Still mix between Figma, prototyping, Slack feedback loops—always quick iteration.
- Channels like “PD Whip” foster rapid internal sharing and feedback.
Systems Thinking: Pitfalls & Challenges
[31:27–33:17]
- The balance: Launching experimental features vs. building a cohesive system.
- “What does not good systems thinking look like? Honestly, I think it's just if you're a little bit too blinders on... not totally understanding that we have a bunch of other things going on that are actually pretty similar.” (32:36)
Designing a Dynamic Interface Library—For Humans & Models
[33:17–34:18]
- Challenge: Build UI components that render natively everywhere, are interactive, and are “AGI-pilled.”
- The model should eventually help compose interfaces—blurring lines between design, engineering, and AI reasoning.
- “...components that stack up to a system that a designer can use, an engineer can use, and the model can use.” (34:11)
The Future Role of a Designer
[34:18–35:48]
- Analogy: Photography didn’t eliminate the need for artistic skill—it democratized creation and raised the importance of curation/editorship.
- “Now anybody can write software for anything... just because anybody can do it doesn't necessarily mean it's going to be good... your job's going to be more and more about helping edit and direct and curate.” (34:39)
Tensions: Designing for Now vs. Designing for the Future
[35:48–40:19]
- Concept of “capability gap”—models can do more than most users know.
- Current limitations (tokens, context windows, prompt engineering) will become archaic quickly; always a moving target.
- “You have to kind of find the balance and that's where like really understanding what it's capable of now and then also like being able to push it and say, well, we actually do want to be able to do this...” (39:51)
Leadership: Fostering Risk and Vision
[40:19–40:58]
- Encourages team to balance exploration of the future with execution for the present.
- “I think that's the big question is finding when the right time is to do that... you have to be able to be very fluid.” (40:42)
The Reality of Working at OpenAI
[41:14–41:55]
- Fast-paced, perpetually evolving, collaborative.
- “Things are changing underneath your feet all day long. And it's very exciting. It's really fun to be like, I don't know, we're going to figure this out as we go.” (41:46)
Balancing Tech-Led and User-Led Design
[41:55–43:12]
- Great designers today must blend responding to tech and anticipating user needs.
- “You have to think not just about the problems, the user problems... but you have to think about what the technology can't do that it should be able to do.” (42:29)
What OpenAI Looks For When Hiring Designers
[43:12–44:51]
- Curiosity, flexibility, fundamentals of product design.
- Candidates don’t need AI backgrounds, but benefit from having experimented with new tools.
- “I've always respected people that will go deep on some side project or get really passionate about some idea... just being truly curious...” (43:45)
Notable Quotes & Memorable Moments
- “A lot of thinking about the model as the product... hoping designers are spending less time in Figma and more thinking about how you interact with this thing.” (09:12)
- “There's a lot of bottoms up... anybody can build an idea now. You can, of course, just like fire up Codex or whatever tool and ship it.” (17:51)
- “The job of a designer is not going away anytime soon... people can recognize good software from bad software. It's not just about does it work, but how it works and how it feels and is it solving the right problems for me.” (35:38)
- “Some people might assume we're like two years ahead... but we're running very close to mostly with where all of these advancements are going. Things are changing underneath your feet all day long.” (41:14, repeated at end as bookend)
Episode Timestamps & Key Segments
- 00:51–05:10 — Ian’s background and journey to OpenAI.
- 06:53–09:38 — Differences in design culture at a research lab like OpenAI.
- 09:38–12:00 — “Designing with the model”: Beyond pixels.
- 12:28–14:51 — UI vs. model paradigm: Direct manipulation and writing tools.
- 14:54–16:49 — Systems thinking: Designing for extensibility and composability.
- 16:49–18:35 — How ideas become shipped features at OpenAI.
- 22:40–26:28 — Evolution of the designer skill set.
- 32:36–34:11 — Pitfalls in systems thinking.
- 34:18–35:48 — The rise of the editor/curator in design.
- 41:14–41:55 — Day-to-day pace/culture at OpenAI.
- 43:32–44:51 — What OpenAI looks for when hiring designers.
Summary Takeaways
- Design at OpenAI is distinctively fast-paced, research-driven, and systemically focused.
- The boundaries between design, engineering, and product management are blurring, with AI unlocking new modes of prototyping and experimentation.
- The most successful designers balance traditional craft with curiosity and adaptability to new tools and workflows.
- Systems thinking is pivotal—both for building flexible products and empowering the model itself to generate and reason about interfaces.
- The future role of designers will center more on editing, curating, and guiding, as the “material” of software becomes increasingly democratized.
For designers, technologists, and anyone curious about the evolving intersection of design and AI, this episode offers both inspiration and a candid look at the tumultuous, dynamic, and rewarding practice of designing on the frontier.
