Microsoft engineer: 'Typing code is absolutely over' as AI agents transform development
Microsoft engineer David Fowler says typing code is over; AI agents take over coding, but humans still needed for oversight.
Fowler's post ultimately implies that developers' roles are evolving, rather than disappearing. Developers are increasingly expected to prompt AI agents, verify outputs and fine-tune results instead of writing code from the ground up. Microsoft-owned GitHub Copilot agentic tools now handle much more of the workflow, including building software, running tests and checking for failures before the project is handed to a human for review. Microsoft is also making its Aspire development platform more agent-friendly, with version 13.2 introducing support for MCP and an agent-friendly CLI.
TechRadar notes that Microsoft CEO Satya Nadella said in 2025 that up to 30 percent of Microsoft's code was already AI-generated, a figure likely much higher today. However, Microsoft's own researchers found that accuracy declines even for frontier models the longer a task runs. CodeRabbit found that AI-generated code contains on average 1.4 times more critical issues and 1.7 times more major issues than equivalent human-generated code. Veracode warned that 45 percent of AI-generated code contains security flaws despite appearing production-ready.
While models are expected to improve with each generation, mistakes will probably remain. So even if typing lines of code is ending, the human developer's role certainly is not.