Dive Club 🤿 — Ep: “Designing Frontier Interfaces at Neuralink and Apple”

Guest: Rooz Mahdavian (Design Engineer, Neuralink; formerly Apple)
Host: Ridd
Date: October 3, 2025

Overview

In this riveting Deep Dive, Ridd sits down with Rooz Mahdavian to explore the bleeding edge of interface design: neural interfaces. From early inspirations involving FMRI movies to Apple Watch Faces and on to pioneering everyday brain-computer interaction (BCI) at Neuralink, Rooz shares rare insight into inventing UX for users who navigate their world in fundamentally new ways. The episode is a masterclass in first-principles design thinking where constraints, empathy, and ambition converge to reimagine computing.

Key Themes & Discussion Points

1. What Is a Frontier Interface?

[00:00] — [07:26]

• Ridd opens by spotlighting the contrast between B2B SaaS and truly novel interfaces: “We talk a lot about craft in B2B SaaS, but what's it like designing frontier interfaces?”
• Rooz defines “frontier interface” as inventing for unknown interaction models — designing for intent beyond mouse, touch, even hands:
  • “Historically, the interaction model…is usually a function of just what the input mechanism is…If you have a neural interface...you could potentially read [intent] directly.” ([07:26], Rooz)

2. Rooz’s Origin Story: From FMRI Dreams to Apple & Neuralink

[01:23] — [05:16]

• Early fascination grew from Berkeley research reconstructing movie frames from FMRI data — “a movie of what they're thinking...it was hard to let go of, it was just unbelievable.” ([01:23], Rooz)
• Making movies as a teen: “the idea that you have these images in your mind that you can now show other people directly was like, really, really magical to me.”
• First design internship: Apple Watch Faces team post-launch. His project evolved into the proactive Siri Watch Face — an interface “that sort of fades into the background and…can hopefully just do useful things for you.” ([01:23], Rooz)
• Neuralink “frontier” epiphany after seeing BCI demos in 2019: “I made the jump from one frontier interface on watch to another. In this case neural interfaces.”

3. Designing for BCI—First Principles and Empathy

[09:10] — [14:13]

• The unique challenge: Designing from complete uncertainty; prototyping user onboarding for Neuralink, calibrating implants “with a blank canvas.”
• The empathy gap: “You're not just imagining a job that you've never had before, [but] a state of being that you've never really come close to.” ([10:17], Ridd)
• Rooz: “It is not something that you can imagine well enough for that to be useful...Closed loop is extremely hard because you ultimately cannot experience the neural interface directly.” ([10:46], Rooz)
  • Open loop: User imagines movement, but doesn’t see real feedback.
  • Closed loop: User observes computer reacting to neural intent—impossible to personally experience without paralysis or an implant.

4. Reinventing the Cursor as a Neural Extension

[14:13] — [20:34]

• The cursor becomes the core 2D expression of intent, but users lack all the tactile/sensory feedback of traditional input:
  • “There's such a rich spectrum of sensory experience that just goes into using a cursor…at this point is so subconscious you don't really think about it.” ([14:30], Rooz)
• First version: A MacOS cursor visually “takes over” and morphs based on neural click probabilities (color/intensity/perspective changes) to convey how close the participant is to clicking, making feedback continuous and learnable:
  • “The intuition was that that would just feel a lot better visually than just like the click happening would, with no feedback at all.” ([14:30], Rooz)
  • “Imagine like if you had a new arm today…the first things you do would be super weird and uncoordinated. But over time, you would learn…” ([14:30], Rooz)

5. Lessons from Real Participants: Evolution of Interaction

[20:25] — [25:57]

• User behaviors forced the team to evolve beyond left/right click to richer actions: scrolling, dragging, zooming—each requiring unique feedback mapping.
• Dwell-based clicking (pausing cursor motion to trigger a click) helps when click decoding is less reliable than motion.
• The cursor UI evolved from pointer to circular “reticle” for better feedback, enabling easier access to interaction “modes,” e.g. shoot cursor right to enter drag or scroll:
  • “The world of BCI is an interesting one…if the model is perfect, it's always going to do what you want…what's unique to neural pointing is we can read some form of our underlying intent…” ([26:08], Rooz)

6. The Ladder of Capability—And Deleting the Cursor

[25:57] — [31:30]

• Rooz’s “ladder” metaphor: stepwise progress from basic pointer parity (“using a cursor to do all this stuff today”) to radical new models.
• The dream rung: “Can we delete the cursor?” — could the system infer intent for most actions without any intermediary.
  • “I'm personally very excited about a world in which you don't need to switch modes…that the model can actually learn…what you want to do.” ([26:08], Rooz)
• Early neural cursor users improved dramatically — from 2bps to 9.5bps (bits per second, a measure of speed/precision), with later participants surpassing able-bodied norms.

7. Behavior-Driven Iteration—Parking Spot & ‘Simulated Gravity’

[31:30] — [36:34]

• Voice input fallback proved invaluable, until ALS users who couldn’t speak arrived — prompting new forms of control.
• The “Parking Spot” was invented: users can ‘yeet’ the cursor into a special zone to easily park/disable it without voice.
  • “...the final thing we landed on was this thing called the parking spot where they could sort of yeet their cursor into the bottom right…then they can actually use a gesture within that surface to bring it back out.” ([31:46], Rooz)
• Simulated gravity: Parking spot uses a virtual hill—a velocity-based nudge—to keep the cursor docked, requiring focus/force to extract, tuned per user ability, with gesture overrides for eye-tracker users.

8. Obsessive Detailing and the Tension of Deletion

[36:34] — [38:03]

• Rooz reflects on “obsessing over the smallest pieces of these interactions” knowing the end goal is to “delete all the work you’ve done” once direct intent can be read.
  • “I think ultimately the goal is to just build an incredible experience…and build towards a world where we could, let’s say, delete the cursor.” ([36:59], Rooz)
• Ridd: “The measuring stick is no longer what a normal person can do on a computer. You know, you’ve blown the roof off of what is possible in terms of interaction…” ([37:20], Ridd)

9. Sci-Fi Futures: Direct Translation of Inner Visualization & Blindsight

[38:03] — [42:54]

• On the far horizon: shrinking the gap from “vibe coding” to “a frame”—instant, model-generated artifacts conjured from thought.
  • “You can just sit down at a computer and daydream with it, and things will pop up on screen that are a direct sort of extension of what you have in your head.” ([38:13], Rooz)
• Blindsight Project: Give sight to the blind by stimulating visual cortex via Neuralink; current fidelity likened to “Atari, not PlayStation 5.”
  • “That’s its own completely distinct design space…how do you recreate an image that is true to life, quote unquote, in this domain?” ([38:13], Rooz)
• Design choices echo old-school graphical tricks like dithering, appropriated for brain stimulation.

10. The Human Impact

[42:17] — [43:28]

• Ridd notes the life-changing real stories: participants regaining the ability to communicate, work, live more independently.
  • “What you all are doing is just about as inspiring as it gets.” ([42:17], Ridd)
• Rooz likens early BCI users to “neural-nauts”: “...10 plus folks who are actually using the thing…it's a completely new space that we're exploring together.” ([42:54], Rooz)

11. Who Thrives as a Designer at Neuralink?

[43:28] — [54:28]

• End-to-end designers/engineers wanted: invention, fast iteration, productizing experiments “you can live with for a month.”
• Not copy-pasting from Mobin, but sweating every detail from scratch — must make even temporary prototypes “amazing” to get a strong signal.
  • “Even for something that you know that you’re probably going to delete, there’s a lot of extra work…just making that amazing.” ([46:39], Rooz)
• “The tightest interplay is really between the machine learning side and the actual interface side of the work…what the models can do defines what the interface can do.” ([46:39], Rooz)
• MVP thinking alone is insufficient; aim for “magical, beyond the local minima of that which works.”

12. A Day in the Life: Sweating the Body Mapping Task

[49:24] — [54:28]

• Example: “Body Mapping Task.” Render a 3D arm participants imagine moving—even animate text cues to reinforce the action (e.g., kinetic type for “squeeze”).
• “If there is this mind body-connection that we can induce by then just having this arm on screen...maybe we can actually have this simple imaginary space…” ([49:47], Rooz)
• The outcome: the hunch didn’t pan out technically, but the ambition and attention to crafted interaction design stands as the blueprint.

Notable Quotes & Memorable Moments

• On first encounter with neural interfaces
  • “The idea that you have these images in your mind that you can now show other people directly was like, really, really magical to me.” – Rooz ([01:23])
• On design empathy for BCI users
  • “It is not something that you can imagine well enough for that to be useful...closed loop is extremely hard because you ultimately cannot experience the neural interface directly.” – Rooz ([10:46])
• On the ambition behind every detail
  • “Even for something that you know that you’re probably going to delete, there’s a lot of extra work that goes into just making that amazing.” – Rooz ([46:39])
• On the ultimate vision of BCI
  • “You can just sit down at a computer and basically daydream with it, and things will pop up on screen that are a direct sort of extension of what you have in your head.” – Rooz ([38:13])
• On the inspiration of participant stories
  • “What you all are doing is just about as inspiring as it gets.” – Ridd ([42:17])
• On naming the first users
  • “We call them, like, the neural-nauts, because like astronauts, it's a completely new space we’re exploring together.” – Rooz ([42:54])

Timestamps for Key Segments

• 00:00–01:23 — Frontier interfaces and the BCI paradigm shift
• 01:23–05:16 — Rooz’s journey: FMRI inspiration, Apple, Neuralink
• 09:10–14:13 — Early BCI prototyping, empathy challenges
• 14:13–20:34 — Redesigning the cursor for neural control
• 20:34–25:57 — Learning from participants and evolving UI
• 26:08–31:30 — The ladder of cursor capability, dreams of ‘deleting’ the cursor
• 31:46–36:34 — Parking Spot and simulated gravity as workaround UI
• 38:13–42:17 — Sci-fi future of direct visual-articulation, ‘blindsight’ for the visually impaired
• 43:48–49:47 — The kind of designer/engineer that thrives at Neuralink
• 49:47–54:28 — Example: The body mapping experiment—ambition, risk, and crafted delight

Final Takeaways

This episode is a masterclass in what it means to design at the edge of possibility:

• True first-principles invention transcends empathy, requiring relentless iteration, humility, and courage to build for lives unlike your own.
• Successes in “frontier interfaces” come from sweating details, embracing failure, and maintaining the ambition to conjure magic—beyond the simply functional.
• Neuralink’s process takes the best from film, art, machine learning, and consumer tech to create new forms of agency—for users who need it most.
• The real north star is not just parity with current tools, but unlocking domains of experience—like expressing dreams directly—that today’s UIs barely hint at.

Whether you’re a designer, engineer, or just fascinated by the future, Rooz’s journey is a powerful reminder: sometimes, the best work is the work you aim to one day delete.