Wasmer's Syrus Akbary on AI-Powered JavaScript Runtime

Wasmer CEO Syrus Akbary reveals how their AI runtime, Codex, drastically accelerates JavaScript development and debugging.

5 min read
Syrus Akbary, Founder & CEO of Wasmer, speaking on camera.
Image credit: StartupHub.ai· OpenAI Youtube

Syrus Akbary, Founder and CEO of Wasmer, discusses the transformative capabilities of their AI-powered JavaScript runtime, Codex. Akbary, a prominent figure in the WebAssembly and JavaScript runtime space, highlights how Codex is fundamentally changing the development lifecycle. The core thesis is that Codex not only speeds up development but also enables continuous, unattended AI assistance, a leap forward from previous AI coding tools.

Syrus Akbary's Vision for AI in Development

Syrus Akbary, as the founder and CEO of Wasmer, brings a deep understanding of runtime environments and the challenges faced by developers. Wasmer is known for its work in sandboxing and running code efficiently across different platforms, often utilizing WebAssembly. Akbary's perspective is rooted in practical development needs, aiming to bridge the gap between AI's potential and the realities of software engineering. His leadership at Wasmer positions him as a key voice in how AI can be integrated into the developer workflow.

The full discussion can be found on OpenAI Youtube's YouTube channel.

Related startups

What Codex Unlocks for Wasmer - OpenAI Youtube
What Codex Unlocks for Wasmer — from OpenAI Youtube

Codex: Accelerating JavaScript Development

Akbary explains that the primary benefit of Codex is its ability to drastically reduce the time required for JavaScript development and debugging. He states, "We were able to create a JavaScript runtime in two weeks, three weeks and a half without code, which would have taken us one year." This statement underscores the immense acceleration Codex provides. It signifies a shift from traditional, time-consuming coding processes to a more agile, AI-driven approach.

The platform is designed to assist not only in writing new code but also in debugging existing code. This dual functionality is critical for modern development, where identifying and fixing errors can often be as challenging as initial development. By automating and streamlining these processes, Codex aims to free up developer time for more complex problem-solving and strategic tasks.

Continuous AI Assistance

A significant advancement highlighted by Akbary is Codex's ability to provide continuous AI assistance. He elaborates on this, saying, "We have been able to let Codex work for a few hours straight without any input whatsoever." This capability moves beyond simple code completion or suggestion. It implies a more autonomous AI agent that can work on tasks, potentially complex ones, for extended periods without human intervention. This is a critical distinction from many AI coding tools that require constant prompting or validation.

This continuous operation allows the AI to tackle larger development segments, such as implementing an entire N-API provider for JavaScriptCore. Akbary notes that this specific task, which is crucial for interfacing JavaScript with native code, was completed in a remarkably short timeframe. The efficiency gained is attributed to the AI's ability to handle the intricacies of the task, including interface changes, workflow adjustments, and test plan creation, all while the development team focuses on guiding the AI's direction rather than writing every line of code.

Shifting Focus: From Code Generation to Guidance

Akbary also articulates a strategic shift in how Wasmer views the role of AI in development. Instead of focusing solely on AI generating code, the emphasis is now on AI guiding the development process. He explains, "What's left for the actual task is to run a SAR-enabled JavaScriptCore... I will drive the build and then run the JSC code against the injected framework." This indicates a move towards a collaborative model where AI handles the repetitive and time-consuming aspects, while human developers provide the strategic direction and final oversight.

This approach is particularly beneficial for languages like Rust, where the team might not be expert C++ developers. Akbary mentions, "The team is really expert in Rust... but for C++ we are not as expert. So, the team is actually expert on Rust, but we are not expert on C++." Codex, in this context, acts as an expert assistant, enabling developers proficient in one language to effectively work with and develop for another, effectively democratizing complex development tasks.

Implications for the Developer Workflow

The implications of Codex for the developer workflow are profound. The ability to achieve complex tasks like creating a JavaScript runtime in a matter of weeks, rather than a year, suggests a future where development cycles are significantly compressed. This acceleration can lead to faster product releases, quicker iteration on features, and more efficient bug fixing.

Furthermore, the shift towards AI guidance rather than direct code generation empowers developers. It allows them to focus on higher-level architectural decisions and problem-solving, while the AI handles the more mundane or complex implementation details. This can lead to more fulfilling and productive work for development teams.

The Future of AI in Runtime Environments

Akbary's discussion of Codex points to a future where AI is deeply integrated into runtime environments. As AI models become more sophisticated, their ability to understand and manipulate code will only grow. This could lead to entirely new ways of building and deploying software, with AI agents playing an increasingly significant role in the development process. Wasmer's work with Codex is a clear indicator of this trend, positioning them at the forefront of this evolution in software development.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.