Did AI kill React Native?

Summary of the Video

1) The “React Native vs native” debate—framed through Shopify’s reversal

• The video revisits an older React Native discussion and updates it with major recent events.
• A key catalyst is Shopify’s announcement that it plans to move off React Native and return to native development (Swift/Kotlin).
• The speaker argues this is disappointing because Shopify’s 2020 bet is credited with boosting React Native’s credibility:
  • Shopify reportedly saw performance wins, at times outperforming native, especially for Android POS systems.
• The video frames Shopify’s shift as both:
  • a technical re-evaluation, and
  • a market shift driven by AI “coding agents” changing the economics of development.
• Core idea: if agents reduce the cost/benefit of “write once” approaches, then some of React Native’s earlier advantages can weaken.

---

2) Why Shopify claims they’re switching back to native

Based on Shopify’s stated rationale (plus the speaker’s interpretation):

• Coding agents reduce “double-build” costs
  • Building separate iOS and Android implementations is said to be less expensive than in 2020.
• Shopify says its earlier React Native advantage was real, but it became less dominant as:
  • model capability improved,
  • agent-assisted development got better at implementing features + testing + review,
  • and native platform tooling advantages remained strong.

---

3) LLMs/agents as a development tool (sponsor + thesis)

The video spends time on an AI coding workflow focused on review and testing:

• Sponsor introduces Graphtile / T-Rex (subtitles have some typos, e.g., “Reptile agent”).
• The approach emphasizes running code in sandboxes to catch problems static review misses.
• The system described (“Reptile agent” plus sub-agents running in separate sandboxes) can:
  • execute and test code,
  • collect artifacts (e.g., images/videos),
  • and provide evidence-based context for fixes.
• The speaker ties this back to Shopify’s story: agents that can actually test and iterate can change mobile development economics.

---

4) Shopify’s specific migration approach + migration risk controls

Rebuild strategy

• Shopify considered:
  • Brownfield migration (incremental change)
  • vs Greenfield migration (rebuild)
• They reportedly conclude greenfield is the “real winner,” likely because agents reduce rewrite/verification cost and enable faster prototyping.

Phased migration timeline (examples)

• Shop app is positioned as the first major migration:
  • ~12 weeks from proof of concept to fully rebuilt and App Store release
  • Mentioned scope includes ~300 screens, widgets, Apple Watch app, complications, and features like Siri Shortcuts.
• Other migrations (e.g., the Shopify app) are said to ship later that year.

Preventing “slop” (bad code output)

• The speaker describes Shopify’s internal system “Helix”, an iterative checkpoint loop where each stage must pass:
  • tests that match behavior in the running app,
  • visual review,
  • “adversarial” code review,
  • plus human approval before proceeding.
• The point: avoid one-shot AI implementations that produce code that becomes unmaintainable or incorrect.

Faster agent feedback loops to avoid slow simulators

• Mobile simulators are described as slow and brittle for agent workflows.
• React Native hot module reloading helps, but doesn’t eliminate simulator control bottlenecks.
• Shopify’s solution is agent tooling that can iterate headlessly by:
  • decoupling business logic from UI so logic runs on desktop,
  • providing a CLI for state inspection and actions without driving UI through accessibility trees/screenshots,
  • only using simulators in remote mode when necessary (UI driven by commands).

---

5) React Native “non-performance” pros/cons the speaker emphasizes

Strengths (as framed by the speaker)

• OTAA (Over-the-Air Updates)
  • React Native can support OTA-style updates by swapping JS bundles without rebuilding the full native app.
  • The video explains that Apple/Google constraints limit what can be reviewed/changed in new features, positioning OTA as important for:
    • turning fixes/features on/off quickly,
    • avoiding long rollout delays,
    • reducing support burden from older installed versions.
• React Native as a “command layer”
  • Not another app platform per se, but a recipe distribution model:
    • JS dictates behavior and what native components render,
    • while the underlying native runtime remains.

Negatives (as framed by the speaker)

• Library bloat / native packaging pain
  • Critiques are mostly aimed at iOS/Android integration tooling (Xcode/CocoaPods/Swift/C++ bridging), not at JS itself.
• “Foot guns”
  • Easier adoption leads to incompetent usage patterns that can cause severe issues in mobile contexts.
• Moving target / churn
  • Large rearchitectures and changes can make integrations harder and version upgrades painful.
• Meta dependency
  • The speaker argues it’s frustrating when native capabilities exist but you must wait for Meta’s support/shims.

---

6) Expo and why it mattered to the Shopify story

• The speaker claims Shopify did not heavily adopt Expo.
• Why this matters:
  • Expo provides build tooling, OTA-style updates, simulator support, and web support.
  • If you aren’t using Expo, you may need to rebuild similar tooling, reducing some ecosystem advantages.

---

7) React Native “new architecture” (performance-related but also tooling)

• The video covers React Native’s new architecture, described as improving:
  • bridging,
  • layout/effects,
  • and synchronization.
• The speaker suggests Shopify’s performance comparisons may be less informative without accounting for new architecture improvements.
• The new architecture is portrayed as important infrastructure (the speaker claims it powered all production apps at Meta).

---

8) Personal experimentation: AI-assisted porting and navigation differences

AI-assisted porting anecdotes

• The speaker describes trying to port an app from:
  • React Native → Swift/AppKit/SwiftUI
• Observations:
  • Agents port effectively when given:
    • an existing codebase,
    • a working implementation,
    • and clear guidance.
  • Simulator-driven iteration remains painful, but automation loops can reduce time-to-port.

Gesture/navigation behavior differences

• The speaker compares navigation/gesture behavior:
  • SwiftUI navigation gestures may not match expected “native gesture areas.”
• The claim: achieving true UX consistency can require rebuilding navigation logic, not just relying on platform defaults.

---

9) React Native team “state” concern: staffing, talent density, and maintenance

A major closing theme is the future of React Native itself, beyond Shopify’s decision:

• The speaker claims Meta is disbanding/downsizing React/React Native teams.
• Consequences described:
  • fewer human maintainers,
  • less thorough evolution,
  • quality/speed plateauing,
  • lower confidence in long-term innovation.
• The speaker contrasts past high-density competence with a smaller team where agents may maintain things less “revolutionarily.”

---

Main speakers / sources (as identified in the subtitles)

• Primary speaker (video host): Jamon (reacting as the main narrator/analyst)
• Shopify author(s)/executives referenced:
  • Mustafa Ali (author of a Shopify article reflecting on 5 years using React Native)
  • Toby (Shopify founder/CEO referenced regarding internal decision-making)
• React Native ecosystem contributors referenced:
  • William (React Native Skia creator/maintainer; also a YouTuber; mentioned as working at Shopify)
  • Jay / “Legend List” creator (performance-focused React/React Native library author)
  • Infinite Red / Jamon’s affiliation (OG React Native context)
  • Software Mansion (React Native integration work; also mentions Reanimated as being built by them)
• Sponsor/source in the middle of the video:
  • Graphtile / T-Rex sponsor (subtitles mention “Reptile agent” and related tooling)