Building ambitious software — Jonathan Kelley, Dioxus Labs & Cognition

Jonathan Kelley, founder of Dioxus Labs, shares insights from building a cross-platform Rust app framework for five years. He details Dioxus's evolution, its core technologies like Blitz and Subsecond, and the transformative impact of AI coding agents. Kelley explains how AI excels at handling Rust's complexity, automating mundane tasks, and serving as patient specialists for knowledge-intensive problems, yet falls short in writing effective tests. He concludes that while AI makes code cheap, quality remains paramount, shifting the engineer's focus to architecture and strategic system design.

2 min read
Building ambitious software — Jonathan Kelley, Dioxus Labs & Cognition
source: https://www.youtube.com/watch?v=H7vFrcNWXzs

The Dioxus Journey: Ambitious Software and the Age of AI Coding Agents

Jonathan Kelley, founder of Dioxus Labs, shares his team’s journey of building an ambitious software project and the profound impact of AI coding agents on their development process.

Inception and Evolution of Dioxus

Kelley initiated Dioxus five years ago as an undergraduate, driven by the vision of creating a cross-platform app framework using Rust, HTML, and CSS. His motivation stemmed from Rust’s native performance, robust type system, and straightforward cross-compilation, aiming to overcome limitations prevalent in existing solutions like React Native and Flutter. The early years were marked by significant challenges, as the team had to build almost every component from scratch—including reactivity systems, font rendering, hot reloading, and application bundling.

Today (as of the presented 2026 perspective), Dioxus has far surpassed its initial goals, boasting nearly 37,000 GitHub stars and an estimated 200 million cumulative end-users. It enables the creation of full-stack web, iOS, and Android applications from a single Rust codebase, deployed in diverse applications ranging from AI assistants and voting software to satellite collision avoidance systems.

Core Dioxus Technologies

Two pivotal technologies underscore Dioxus’s capabilities:

  • Blitz Rendering Engine: A lightweight yet fully-featured HTML and CSS rendering engine. It integrates a browser-grade CSS engine extracted from Firefox, a custom HTML DOM, and a hybrid GPU rendering pipeline. Blitz applications are remarkably efficient, typically under 5MB in bundle size and consuming less than 50MB of RAM, in stark contrast to more resource-intensive frameworks like Electron.
  • Subsecond Hot Reload Engine: A unique, generic hot reload engine designed for Rust, C, and C++. This technology allows live patching of running native applications in approximately 100 milliseconds, showcasing wide language and runtime support, including WebAssembly. Kelley notes that achieving this level of broad support for native compiled code hot reloading was an unprecedented technical feat.

The Impact of AI Coding Agents

Initially skeptical, the Dioxus team experienced a significant shift when AI coding agents demonstrated exceptional proficiency in Rust. This newfound capability led to an initial

Share post

Stay In The Loop! Subscribe to Our Newsletter.

Get updates straight to your inbox. No spam, just useful content.

Recommended for You

Code Quality in the Age of AI: Why Great Code Isn't Enough

Code Quality in the Age of AI: Why Great Code Isn't Enough

Meenakshi Kodati discusses how AI is transforming software engineering, shifting the focus from coding to critical decision-making. As AI rapidly generates code, the true challenge and value lie in architectural choices, system-level understanding, governance, testing, and continuous quality assurance to deliver meaningful business outcomes, emphasizing human judgment over mere implementation.

How to Get Your Org to Adopt Coding Agents (Without Shipping Garbage) — Eyal Blum, Figma

How to Get Your Org to Adopt Coding Agents (Without Shipping Garbage) — Eyal Blum, Figma

Figma's internal AI agent adoption journey faces challenges like reduced developer agency, skepticism from senior engineers, and communication inefficiency. Solutions include investing in verification, using a testing pyramid for agent review, prioritizing detailed planning over prompting, engaging skeptics to build AI safety roadmaps, and implementing attention-aware communication by clearly marking AI-generated content.