Code Workflows in the Age of Agents
twiecki · x · 2026-07-18
This post relays a perspective on AI code workflows: when agents generate code faster than humans can review it, PRs become a bottleneck. In the future, code could be pushed directly to the trunk, with agents automatically summarizing the changes.
A core takeaway: Models shouldn't decide when a task is complete; deterministic checks must be placed where the model cannot access them. Meanwhile, Git will likely endure, though whether GitHub remains the dominant platform is uncertain. Another guest adds from an evaluation standpoint: agent = model + harness. When evaluating, you must treat the entire system as the subject, not just the model itself.
More from coding & agent
- Devin adds e2b sandboxes for remote agent execution — badphilosopher · 2026-07-22
- Hermes Agent Refactoring Proposal: Decoupling via Event Bus and Monorepo Slicing — Promptmethus · 2026-07-22
- ty now reads Pydantic config keywords and field metadata — charliermarsh · 2026-07-22
- Pensar Launches AI Security Agent to Autonomously Discover and Patch 0-Days — andriy_mulyar · 2026-07-22
- ty adds first-class Pydantic support, including strict and lax field handling — charliermarsh · 2026-07-22
- Google launches Gemini 3.5 Flash Cyber for CodeMender, with limited access for governments — GoogleAI · 2026-07-22