AI-coding agents are now the default. What comes next?

This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).

Linear went all-in on AI coding agents in January 2026 after a company-wide mandate to use Claude Code or Opencode. The results: 30% more PRs and 33% more issues closed per engineer within a month. Key challenges now include context bloat (addressed via the open-source rtk tool, cutting token usage 40%), rising AI costs (including a $6K Cursor bill from one migration), and code review becoming the new bottleneck. Linear still keeps humans in the loop for approvals and prompt injection checks. Engineering rigor has shifted toward systems thinking, with more RFCs written upfront to feed agents better context. The post argues that fully autonomous deployments are coming but the infrastructure to support them safely isn't ready yet.

8m read timeFrom leaddev.com
Post cover image
Table of contents
Your inbox, upgraded.Getting on board with AI-coding agentsManaging context bloatMore like thisKeeping the budget in checkHuman in the loop, for nowBuilding the system for agentic automation
48 Impressions