Sea's View on the Future of Agentic Software Development with Codex

Sea's View on the Future of Agentic Software Development with Codex

A conversation with David Chen, Co-Founder of Sea and Chief Product Officer at Shopee.

Sea Limited (Sea) is a leading global tech company founded in Singapore, with businesses spanning digital entertainment, e-commerce, and digital financial services. Its engineering teams build and operate products at significant scale across some of the world’s most dynamic markets.

Sea is rolling out Codex across its developer organisation, with our internal data showing 87% of users are weekly active users. For the company, AI-assisted software development is not simply a way to improve productivity at the margins, but a deeper shift in how engineering teams navigate complexity, build resilient systems, and move from ideas to implementation.

We spoke with David Chen, Co-Founder of Sea and Chief Product Officer at Sea's e-commerce business, Shopee, about why the company is making this bet, how AI agents are changing how its developers work, and what AI-native software development could mean for Southeast Asia and the broader Asia Pacific region.

At Sea’s scale, engineering isn't just about writing code. It’s about managing large-scale systemic complexity across fragmented, hyper-localized markets. We see the ongoing developments in AI leading to a fundamental shift in how software is created and how our engineering teams operate at scale.

Agentic AI coding tools like Codex are not just about improving localised productivity. They represent a structural multiplier that can help our engineering organisation ramp up speed, responsiveness, and effectiveness amid an increasingly complex operating environment.

What stood out was its ability to go beyond autocomplete and provide deep contextual awareness of our large and disparate codebases. In a massive microservices architecture, the friction is not typing syntax. It is tracing dependencies, understanding legacy logic, and maintaining reliability under peak loads.

Codex acts as a localised knowledge engine, drastically reducing the time it takes an engineer to navigate unfamiliar services and allowing our teams to shift their cognitive load to higher-level tasks such as architectural design and product innovation.

It has been encouraging to see the Codex adoption trends among our developers, particularly among frequent users, with many citing improvements in experimentation speed and development workflows. Based on internal feedback among developers who rated Codex 4 or 5 out of 5, 73% said they would recommend it to colleagues.

The most profound shift is realizing our developers are using Codex to 'think better', not just type faster. We are actively transitioning from using AI as a passive autocomplete mechanism to integrated agentic workflows.

In practical terms, this means AI agents are increasingly operating within our CI/CD pipelines—reasoning through product requirements, autonomously proposing test-driven implementations, surfacing edge cases in distributed systems, and accelerating debugging loops.

Many assume AI simply increases velocity. But at Sea, we are also using it to drive engineering discipline. By allowing AI to rapidly prototype alternative implementations and generate exhaustive test coverage, we are moving faster and are systematically paying down technical debt and shipping more resilient systems.

If you look at past technology revolutions, Southeast Asia has consistently leapfrogged traditional technology adoption cycles, such as the move directly to mobile-first and super-app ecosystems. Because developers here must solve highly complex, multilingual problems across fragmented commerce, payment, logistics, and communication networks, Southeast Asia is the perfect proving ground for AI-native software development.

Looking ahead, I foresee a fundamental reconfiguring of engineering teams. Software teams will become increasingly more leveraged as AI agents take on more operational execution work. As agents abstract away the implementation layer, the 'developer' evolves into a 'system orchestrator' who spends the bulk of their time on tasks such as product judgment, system design, and orchestrating AI-driven workflows.

At the same time, development cycles will likely become more iterative and continuous as the cost of experimentation and execution continues to fall.

For technology leaders, I would say that this is not simply a tooling upgrade; it is an organizational paradigm shift. The winners will be those who relentlessly redesign their engineering culture and workflows around human-AI collaboration today, rather than bolting it onto legacy processes tomorrow.

Southeast Asia has an incredibly vibrant builder ecosystem, but the tooling gap has historically constrained execution speed. Internally, we have seen how AI-assisted workflows can help engineers learn faster, experiment more freely, and move from ideas to implementation more efficiently.

By bringing this Codex Hackathon Series to the broader developer community outside the company, we are democratizing access to the world's most advanced AI primitives. Doing so can drastically lower the barrier to entry for local developers, enabling them to move from raw curiosity to deploying scalable, AI-native applications in a matter of hours.

This is about building a compounding AI-native talent ecosystem. By upskilling the region’s developers today, we are collectively accelerating Southeast Asia’s trajectory as a global hub for AI-driven innovation.