acquisitions

Cognition Adds Dioxus Team to Advance Devin Coding Agent

mm
Add Unite.AI to your preferred sources on Google

Cognition announced on September 10, 2026, that Jonathan Kelley and the entire Dioxus team are joining the company to work on Devin, its autonomous software engineering agent, while continuing to build and maintain the Dioxus family’s open-source projects.

Dioxus Labs published its own announcement the same day, stating that the team is joining Cognition to accelerate the development of Devin, which it described as Cognition’s autonomous cloud coding agent. Dioxus is an open-source Rust framework for building full-stack web, desktop, and mobile applications from a single codebase, according to the project’s official site.

On its website, Cognition describes Devin as the first autonomous software engineer, a system that plans, writes, tests, and ships production code on its own while working inside a customer’s codebase and the tools the customer’s team already uses. The company says Devin is deployed at some of the largest and most complex institutions in the world. In its announcement, Cognition said it used Dioxus extensively to build and improve the performance of Devin, and that when the two teams met, it recognized a natural fit.

Devin Work and Open-Source Plans

According to Cognition’s announcement, the Dioxus team will bring expertise in app development and systems engineering to Devin, with work focused on improving Devin’s virtual machine, computer use skills, and testing capabilities.

Cognition said that, as one team, it plans to continue supporting Dioxus, Blitz, Taffy, and Subsecond while increasing investments in Dioxus-Native and Blitz. Dioxus Labs likewise said the team will keep working on Dioxus, Blitz, Taffy, and Subsecond and will maintain its wider array of open-source projects for the benefit of the community. Kelley wrote that Nico Burns, the team’s lead engineer on Blitz, will be devoted full time to maintaining and improving Dioxus, and that Burns’s productivity has increased substantially since he joined Cognition, with development of Blitz, Taffy, and Parley accelerating.

Kelley wrote that the team will invest heavily in Dioxus-Native and Blitz, its native HTML/CSS renderer. He wrote that while the team remains excited about Dioxus itself, technologies like Blitz, Subsecond, wasm-split, and DX will do more in the hands of everyone rather than only developers who use Dioxus, pointing to Cloudflare, which he said recently used Blitz as the core engine of its new agentic web browser.

How the Teams Connected

Kelley’s post traces the relationship to late 2025, when the team noticed that LLM code generation had become very good at writing Rust, an improvement he suggested stemmed from advances in reinforcement learning. He wrote that Dioxus saw the shift in its download numbers as more users picked up Rust and Dioxus to build their apps, and that the biggest request from users was, and remains, a more comprehensive ecosystem of Rust crates for capabilities like camera, Bluetooth, and storage.

Given the surge in Dioxus use, the team’s own increased AI-assisted ambition, and the scale of operating-system-specific work on its roadmap, Dioxus Labs decided to build its own cloud coding harness, called SkyVM. Kelley wrote that SkyVM was engineered to support all major operating systems (macOS and iOS, Android, all Linux variants, and Windows) and that the team hand-rolled new virtualization technologies to achieve secure virtual-machine snapshotting, forking, and rollbacks, allowing new virtual machines to spawn and resume in under 50 milliseconds.

At the same time, Cognition was shipping its new Devin CLI, and the Dioxus team learned that the Devin CLI’s terminal renderer is built with Dioxus, a rebuild of the framework’s old, deprecated TUI renderer. Kelley wrote that the team was invited to the Cognition office for a demo, and that it became immediately clear that both teams shared a vision for the future.

From Side Project to Venture-Backed Startup

Kelley wrote that Dioxus began as a side project to pass time during the last summer of Covid, then grew into a budding open-source project, then a full-time job, and then a venture-backed startup. The project’s blog archive dates his move to full-time work on Dioxus to May 5, 2023. He wrote that the team collectively raised $3.5 million in venture funding, and he closed the announcement by thanking investors at Y Combinator and Khosla Ventures, champions at Futurewei, and the community the team built over the years.

Kelley wrote that many people had wondered how the team would eventually monetize Dioxus, and that joining Cognition gives the team the opportunity to continue developing Dioxus and Blitz without financial pressure while helping to shape the future of software engineering with Devin. He wrote that the team can now take many of the ideas it had for Dioxus (simplified cross-platform app development, rapid development speeds, and a reimagined software deployment system) and build them for all developers.

The Projects Named in the Announcements

The Dioxus 0.7 release notes, dated September 8, 2025, describe the work behind Subsecond, Dioxus Native, Blitz, wasm-split, and the DX command-line interface. According to the notes, Subsecond hot-patches Rust code at runtime across web, desktop, and mobile platforms, letting developers edit code and see changes without losing their app’s state. Dioxus Native is a WGPU-based renderer that paints Dioxus apps entirely on the GPU without a webview, and Blitz is a modular HTML/CSS rendering engine built on the team’s Taffy flexbox layout engine, the Stylo CSS resolution engine shared by Firefox and Servo, and Linebender’s Vello GPU compute renderer.

The notes describe wasm-split as adding automatic code splitting and lazy loading for WebAssembly apps, and the DX tooling as usable with any Rust project rather than only Dioxus projects.

Cognition said that, powered by Devin, it expects even more improvements in the Dioxus platform in the months to come.

Jonas Reeve is an AI-generated analyst at Unite.AI, focusing on cognitive AI, artificial general intelligence (AGI), and the theoretical foundations of machine intelligence. His work explores how learning, reasoning, memory, and abstraction emerge in both biological and artificial systems, drawing connections between modern AI architectures and long-standing questions in cognitive science and philosophy of mind.

With a conceptual and reflective approach, Jonas examines frameworks such as reasoning models, agentic systems, emergent cognition, and alignment theory, aiming to clarify what progress toward AGI actually means—and what it does not. Rather than chasing timelines or hype, he emphasizes first principles, conceptual rigor, and the limits of current models.

Articles authored by Jonas Reeve are AI-generated and reviewed by Unite.AI’s editorial team to ensure accuracy, clarity, and responsible discussion of advanced AI concepts.