In a recent podcast Meta CEO Mark Zuckerberg unveiled a sweeping vision for the future of software engineering: one in which AI agents write, test, debug, and optimize most of Meta’s code without human intervention. While autocomplete tools like GitHub Copilot have already become staples in developer workflows, Zuckerberg suggests we’re on the brink of a revolution where AI not only assists, but leads.
“I think sometime in the next 12 to 18 months, we will reach the point where most of the code that goes into these efforts will be written by AI,” he said. “And I don’t mean like autocomplete… I mean, if you give it a goal, it can run tests, it can find issues, it will write higher quality code than an average very good person on the team already.”
Meta’s ambitious plans center around its Llama project, the company’s flagship open-source large language model. According to Zuckerberg, AI agents are already “fully plugged into [Meta’s] tool chain,” and internal teams are building specialized agents—coding and research assistants—with the explicit goal of advancing Llama development.
Coding Agents at Meta: Not Tools, But Teammates
Unlike enterprise-focused tools like Microsoft’s GitHub Copilot, Meta is developing AI agents tailored specifically for internal use, rather than building general-purpose developer software. Zuckerberg emphasized that these tools are purpose-built, designed to optimize workflows inside Meta’s ecosystem.
“We are making [coding agents] for a specific goal. We are not trying to build a general developer tool,” he said. “It is going to end up being an important part of how stuff gets done.”
These agents are designed to understand goals, navigate toolchains, and execute programming tasks with minimal oversight. Importantly, they are built not to replace developers entirely but to elevate productivity, eliminate bottlenecks, and automate routine engineering tasks—while simultaneously pushing the boundaries of Meta’s AI research and deployment.
Zuckerberg has voiced this sentiment before, noting earlier this year that the company would eventually rely on “AI engineers instead of people engineers.” He even claimed AI is now capable of replacing a mid-level software developer.
Industry-Wide Shift: AI Code Generation Becomes the Norm
Zuckerberg’s predictions align with a growing consensus in the AI industry. In March 2025, Anthropic CEO Dario Amodei predicted that AI would generate 90% of code within three to six months, and nearly all code by the end of the year.
Google CEO Sundar Pichai also confirmed that 25% of the company’s code is now written by AI, while OpenAI’s Sam Altman claimed that some companies are already generating 50% of their code with AI.
This rapid acceleration suggests that AI-assisted development is no longer experimental—it’s operational. Across tech giants, AI coding agents are increasingly trusted with core product work, infrastructure development, and even the building of new AI models themselves.
ALSO READ: Call for Cyber Experts: Join FCRF Academy as Trainers and Course Creators
Critics and advocates alike have raised questions:
- What does this mean for the future of software jobs?
- Can AI ensure secure, ethical, and high-quality code at scale?
- And how will engineering teams evolve in a world where code is no longer human-first?