GitHub published a developer guide outlining an 'agentic engineering' workflow that embeds agents across planning, building, verifying and observing. It cites a 400%+ year-over-year jump in weekly commits alongside a 441% longer PR review time for high-AI-adoption teams, arguing the bottleneck has shifted from writing code to review and verification.
Agentic Engineering
智能体工程
Engineering practice of building AI systems that autonomously perceive environments, plan actions, and execute complex tasks
Curated Intelligence
Multiplayer game Spawn demonstrates agentic engineering with frontier models, allowing players to generate and interact with game assets via natural language prompts. The case validates low-latency reasoning and state management of LLMs in complex dynamic environments, offering a viable technical paradigm for AI-native gaming.
GitHub's 2026 Octoverse report reveals that agentic engineering has driven monthly merged pull requests to triple within a year, reaching 220 million, with commits hitting 7.38 billion. This surge signifies AI coding agents have transitioned from auxiliary tools to core drivers of the software development lifecycle.
Anthropic has built a custom source control system to replace Git after AI agents drove daily code merges from 2,500 to over 30,000. The new system utilizes 'Antique Change Requests' instead of traditional Pull Requests to handle high-concurrency versioning challenges. This shift highlights the critical need for re-engineering developer toolchains in the era of agentic coding.
The LangChain team announced enhanced debugging capabilities in LangSmith specifically for 'Deep Agents,' addressing the challenge of tracing errors in multi-step reasoning and tool usage. The update introduces Polly, an AI-assisted debugger that automatically analyzes agent execution traces, decision logic, and intermediate states to help developers quickly identify and fix faults in complex agentic systems.
Zhipu AI has released and open-sourced its flagship model, GLM-5.1, marking a breakthrough in long-horizon tasks by enabling over 8 hours of autonomous coding to deliver complete engineering solutions. The model sets new global records on key benchmarks like SWE-Bench Pro, ranking third overall and first among open-source models.
Zhipu AI has officially launched GLM-5, a next-generation flagship model designed to shift coding paradigms from 'Vibe Coding' to 'Agentic Engineering'. By integrating DeepSeek Sparse Attention (DSA) for cost efficiency and a novel asynchronous reinforcement learning infrastructure, the model achieves state-of-the-art performance in real-world software engineering tasks, rivaling Claude Opus 4.5 and GPT-5.2.