AI Now Writes Most Code at Uber, Engineers Shift to Oversight and System Design
AI is transforming engineering roles as Uber developers move from coding lines to guiding systems, validating outputs, and solving complex problems.
The software development playbook inside global tech firms is being rapidly rewritten, and fresh insights from Uber reveal just how deep that transformation runs. The company’s Chief Technology Officer, Praveen Neppalli Naga, has outlined how artificial intelligence is no longer a supporting tool for engineers but an active creator of production code.
https://x.com/praveenTweets/
In a recent update, Naga said Uber is witnessing large-scale adoption of what he called "agentic software engineering." AI systems are now independently generating substantial volumes of code, often without humans typing a single line.
According to him, Uber’s internal AI coding agent produces nearly 1,800 code changes every week on its own. AI involvement is also widespread across the workforce, with about 95 percent of engineers using AI-powered tools monthly. Even within standard development workflows, close to 70 percent of committed code is now machine-generated.
This shift marks a move far beyond autocomplete and code suggestions. Engineers are increasingly assigning entire tasks to autonomous AI agents that operate in the background. Uber’s internal system has rapidly scaled its contribution—from under 1 percent of code changes to roughly 8 percent within months—without direct human authorship.
Naga described the moment as a "reset moment" for engineering, emphasizing that adoption is being driven organically by developers exploring AI tools rather than through executive mandates.
What engineers are doing now
As machines take on routine coding duties, human engineers at Uber are evolving into higher-level roles. Rather than writing every function and script, developers are concentrating on architecture design, reviewing AI outputs, and ensuring systems function reliably.
Naga explained that engineers are transitioning from passively accepting AI suggestions to actively delegating complex tasks. Their time is now spent directing AI systems, verifying results, and making critical decisions about software structure and scalability.
The company is also embracing "agent-style workflows," where AI tools function more autonomously. Around 84 percent of Uber’s AI users are already working within these frameworks, signaling deeper automation across engineering operations.
Despite the automation surge, human judgment remains central. Engineers continue to define product requirements, collaborate across teams, and ensure AI-generated software aligns with business objectives. The emphasis is shifting from execution-heavy tasks to governance, coordination, and advanced problem-solving.
An industry-wide transition
Uber’s transformation mirrors a broader industry pattern. Several technology leaders have acknowledged that AI is steadily becoming a core contributor to software development.
At Microsoft, CEO Satya Nadella said in 2025 that 20 to 30 percent of repository code was already software-generated. Similarly, Google CEO Sundar Pichai noted that AI accounts for well over 30 percent of new code.
AI-native firm Anthropic has gone even further. Its internal tools reportedly generate nearly all company code. Yet, hiring continues, indicating that skilled engineers remain essential even as responsibilities evolve.
Other industry leaders share this outlook. Jensen Huang and Mark Zuckerberg have both argued that AI should manage repetitive coding work, enabling engineers to focus on innovation, complex systems, and strategic decisions.
As AI matures, the engineer’s keyboard may be quieter—but their role is becoming more influential than ever.
