Thoughtworks Technology Radar Analyzes AI Assistance Trends in 2025

Thoughtworks has published volume 33 of its Technology Radar, highlighting shifts in AI assistance within enterprises. The report notes a move towards 'context engineering' and Model Context Protocol (MCP) while previous trends such as prompt engineering are now less emphasized. This change reflects AI's maturation in software engineering.

The report acknowledges successful AI applications, especially in understanding legacy codebases. However, it also highlights the infrastructure needs and rise of AI-related antipatterns. CTO Rachel Laycock observes a shift from casual approaches like 'vibe coding' to more structured efforts addressing context, infrastructure, and security. The year has been experimental for AI, revealing significant benefits.

The report explores themes such as the integration of MCP for agent awareness, AI infrastructure orchestration, and the emergence of AI coding workflows. It also cautions against AI-induced challenges, urging for human oversight and skepticism to prevent negative impacts.