AI and I — Episode Summary

Podcast: AI and I
Host: Dan Shipper
Episode: Arc Had Millions of Users. Why They Left It Behind for Dia. | Josh Miller and Hursh Agrawal, The Browser Company Cofounders
Date: July 2, 2025

Episode Overview

In this lively, heartfelt conversation, Dan Shipper sits down with Josh Miller and Hursh Agrawal, cofounders of The Browser Company (makers of Arc and now Dia), to trace the tumultuous and visionary journey from pioneering a beloved browser (Arc, with millions of users) to a courageous full reset in the AI era with their new product, Dia. What does it mean to abandon a successful product, withstand public criticism and self-doubt, and bet everything on a seismic platform shift toward “the intelligent, personal, AI-powered browser”? This episode explores the vision, pain, and discovery behind that leap, illuminates the making of Dia, and teases broader lessons on technology, product design, and finding courage as a founder.

Key Themes & Discussion Points

1. The Leap from Arc to Dia: Recognizing the Platform Shift

• The Decision to Pivot (00:00–07:00):

  • Josh and Hirsch reflect on the heady early days after Dia's private beta and admit last year "sucked," despite Arc's traction.
  • They stress that a profound technological shift—the arrival of powerful LLMs—drove the pivot more than “fixing Arc’s growth.”
  • Josh on the excitement:

    "What feels like your computer in five or ten years is actually going to be this layer that sits across all of your devices... You’re turning to it for this personal intelligence layer that is a wild invention of humanity." (00:35, Josh Miller)

• Why Arc Couldn’t Make the Leap (05:15, Hirsch):

  • "ARK had this problem we call the novelty tax: so much new stuff, which attracted early adopters but made the mass market hesitant. We tried so many things to make ARK approachable, then AI started taking off..."
  • Internal inertia: "We loved ARC so much that we were hesitant to change it... We really had to start from scratch."

2. Internal and Community Tensions

• Public Backlash & Internal Struggle (07:08–09:37):
  • The team underestimated the strong negative reaction from Arc fans and underestimated the challenge of getting their own team—who joined for Arc, not “the next thing”—on board.
  • Founders’ self-doubt:

    "It wasn’t just me and him anymore... At every moment, Hirsch and I looked across a table and said, ‘Should we do this?’ But we were running a company with 70 people, millions using the product… Internally, that was a winding journey." (08:20, Josh Miller)

3. The Catalysts: LLMs, New Interfaces, and the “Intent Funnel”

• The Spark (10:04–12:13):

  • The founders describe the spell cast by large language models:

    "You can’t stop thinking about it. You're staying up till 3am. Every bone is screaming this is why we got into software." (10:04, Josh Miller)

    • They started experimenting—first with an “Arc Search” mobile companion.
      • Instead of returning search links, Arc Search would generate a webpage as the direct answer.
      • Its simplicity and explosive user reaction revealed a key design lesson: "Change only one thing at a time; make that one thing what you talk about."

• Strategic Realization — The Browser’s Two Crucial Roles (14:36, Hirsch):

  • Browsers both funnel intent (search) and act as an application platform.
  • AI fundamentally changes both—fulfilling user intent directly, using tools, and integrating context.
  • This alters the role and opportunity for browser software itself.

4. The Messy Path to “We Have to Start Over”

• Indecision and Leadership Lessons (17:24–20:31):

  • Months of indecision, regret over not calling the pivot sooner:

    “It probably wasn’t until September last year that we made the call to say it is a completely new product called something completely different with no connectivity to Ark... That period is the one I regret most.” (17:24, Josh Miller)

  • Conviction about AI was scary—technology was unproven, and “AI browser” sounded like hype.

• Team Culture to the Rescue (23:12, Josh Miller):

  • Company benefited from deep prototyping culture: “Assume you don’t know. Prototype it and see.”
  • The gradual build of capability and confidence allowed staff to come on board slowly, despite initial anxiety.

5. Why Not Just Keep Arc? Taking the Big Swing

• Mission Alignment over Incremental Growth (26:17, Josh Miller):

  • “Why wouldn’t you be OK with a couple million-user browser that makes decent money? ... The motivation was always to build something at that [world-changing] level of ambition, not to make a niche productivity app.”

• The Broader Shift—and Memory as a Central Differentiator (28:39, Hirsch):

  • Pivot was about capturing an AI platform moment, not just building incrementally better browser features.

6. Building Dia: Technical and Product Design Reinvention

• Architectural Overhaul for Speed & Flexibility (33:19–35:46):

  • Learned from Arc’s architectural mistakes—Dia is fast, performant, and prototyping-friendly.
  • “If the strategy was a mass market browser people would use irrespective of the AI features, then it had to be fast…and a really good browser.” (34:13, Hirsch)
  • Used Swift’s concurrency model and custom architecture to achieve both speed and product iteration.

• Design Principle: The Browser is ‘Enabling Tech’—the Main Event is the Personal Intelligence Layer (36:58–39:41):

  • Inspired by the “Internet Computer” vision: an intelligence that sits across all devices, enabling new interactions.

  • "You’re not turning to it for the tabs, or even the desktop part. You’re turning to it for this personal intelligence layer." (38:51, Josh Miller)

7. Skills: A New Layer of “Just-in-Time” Personal Software

• Dia’s “Skills” Platform and Bottom-Up Use Cases (49:53–56:25):

  • Launching and opening up a “skills” layer (mini, AI-driven apps, extensible by users).
  • Surprised by how quickly people grasped the possibilities—creating unique and personal use cases (e.g., automatically fetching guitar chords from YouTube videos).
  • “AI enables a new class of software, where instead of building for a large TAM you build for hyper-personal, niche uses—just in time, as needed.” (56:00, Hirsch)

  “Non-technical people like me want to feel technical... You're inventing a new programming language.” (57:31, Josh Miller, quoting a message to his designer)

8. On Doubt, Resilience, and the Emotional Toll

• Year of Pain, Lessons of Perseverance (60:14–67:42):
  • The aftermath of pivoting from Arc was emotionally brutal: criticism from press, users, and even within the company.
  • Both founders deeply questioned themselves, struggled with employee morale, and weathered personal hardship (Hirsch had a newborn, Josh had two young kids).
  • The power of founder partnership and personal support systems came up again and again:

    “We both have funks and highs… We now know the signs to pull each other out." (63:58, Josh Miller)

  • Emotional resilience and, ultimately, validation:

    "Right now, I'm not this confident that Dia will be the next Chrome, but I know what me and Hirsch felt in our bones was correct." (65:29, Josh Miller)

9. Moats, Competition, and The Importance of “Vibes”

• How Do You Compete Now? (67:42–73:34):

  • The competitive landscape is heating up: perplexity, AI labs, tech giants entering the ‘AI browser’ space.
  • The founders believe traditional moats (distribution, engineering, IP) are shifting—or vanishing.
  • Their new philosophy:

    “It’s all noise… The answer is: we’re going to do our thing. My learning from the last years: when we do our thing, at least some people resonate with it. Vibes is the moat, and we'll figure out the rest." (70:40–71:00, Josh Miller, 71:08, Dan Shipper)

  • Emphasis on execution, product quality, taste, and “emotional intelligence” as differentiators—believing that most competitors are too engineering-driven and miss the personal, subjective, and even artistic experience that truly resonates.

• The Promise of Emotional and Social Intelligence (73:34–75:54):

  • “We're going back to this home on the internet… I’d be surprised if the other AI browsers have that as central as their north star." (73:34, Josh Miller)

10. Inspiration: Romanticism, Culture, and Personal Connection

• Product Design Through the Lens of Romanticism & History (77:04–84:15):
  • Josh and the team intentionally draw inspiration from outside tech—art, history, and especially the Romantic period as a counter to tech’s rationalism:

    “That’s the profound shift: this new relationship with technology and with this intelligence… I want to read about romanticism. I want to understand how in the past, when your relationship with technology changed, how did art and culture react?” (77:04, Josh Miller)

  • LLM-powered software as “vibes-based” and more like artistic material than algorithmic machinery; you can’t fully explain or reduce it to rules.
  • AI as a profound new medium that’s as much about feeling as about function.

Notable Quotes & Moments

• On Memory, Context, and Unprecedented Shifts:

  "Memory... which is core to what we're doing now, I don't think was part of the conversation and now is arguably the most central idea behind Dia." (19:00, Josh Miller)

• On Platform Risk and Betting on the Future:

  "It was a real act of introspection to say, am I disconnected, or do I just see what other people don't because of what we're doing? And for me, it was this constant back and forth in my head." (21:10, Josh Miller)

• On Team, Trust, and Prototype-Centric Culture:

  "One of our saving graces is that we could just sort of frame it as what it was at the time: an inclination, a hunch." (23:12, Josh Miller)

• On Product Vision and Building for the Future:

  "It’s not about the browser; the browser is almost incidental… What you’re turning to it for is this personal intelligence layer that is a wild invention of humanity." (38:51, Josh Miller)

• On Reinvention and Failure:

  “The combination of Hirsch and then my increasing conviction that this was going to change all of software, as bombastic as that sounded… But what happened next was indecision and a lack of excellent leadership, at least on my part.” (17:24, Josh Miller)

• On Emotional Toll:

  "This year, professionally, was by far the heaviest I felt in my body… constant doubt and ups and downs... never-ending." (62:51, Josh Miller)

• On Competition and Moats:

  “Vibes is the moat, and you'll figure out the rest, which…I actually love. I think that's totally right.” (71:08, Dan Shipper) “I just feel like we have a clarity of thought and perspective that feels unique. If we stay locked into being our most energetic selves, that's how we compete." (68:33, Josh Miller)

• On Personalization and a New Kind of Software:

  "AI enables a new class of software where we can build software that allows people to make their niche software just-in-time, whenever they need it." (56:00, Hirsch)

• On the Messy, Nonlinear Path of Innovation:

  “To me, when Abby kind of reminded me of [Romanticism], I just can’t unsee it.” (82:06, Josh Miller)

Selected Segment Timestamps

| Timestamp | Segment | |---------------|-------------------------------------------------| | 00:00 – 05:15 | Opening: Reflection on Arc, the shift to Dia | | 05:15 – 10:04 | Why Arc’s early adopter appeal couldn’t scale | | 10:04 – 12:13 | The “spark” from LLMs and birth of Arc Search | | 14:36 – 17:07 | Rethinking browser as “intent funnel” | | 17:24 – 20:31 | Leadership, regret, and calling the pivot | | 23:09 – 25:05 | Team culture enables risk and adaptation | | 26:17 – 29:23 | Why not settle for millions? World-changing ambition | | 33:19 – 35:46 | Technical overhaul and product requirements | | 36:58 – 39:41 | The “Internet Computer” and personal AI vision | | 49:53 – 56:25 | Release, rapid skill innovation, surprise use cases | | 60:14 – 67:42 | Emotional reality of the “toughest year” | | 68:33 – 73:34 | Facing competition, shifting moats, importance of taste and “vibes”| | 77:04 – 84:15 | Philosophical foundation—romanticism, culture, AI as art/material |

Episode Takeaways

• True platform shifts require courage to walk away from success as well as failure.
• Prototypes beat PowerPoints—most vision can only be made clear by building.
• In the AI age, software moats may favor culture, speed, taste, and “vibes” over distribution or mere technical execution.
• LLMs allow software to become deeply personal, subjective, and open to “mini” creative expression by its users.
• Resilience, strong teams, and support matter as much as vision—the emotional toll is real.
• Great product teams in tech should look for inspiration as seriously in the humanities as in engineering.
• Big changes will seem crazy before they’re obvious. If you’re right, you’ll look crazy right up until you aren’t.

Most Memorable Quote

"Vibes is the moat, and you'll figure out the rest."
— Dan Shipper & Josh Miller (71:08)

This episode is a candid, inspiring narrative about the risks and rewards of ambitious product leadership in the AI era, packed with wisdom for founders, builders, and anyone interested in the future of how we interact with technology.