UseAIHereArticles
English-language articles from UseAIHere: in-depth guides and lecture notes on AI tools and AI-assisted software engineering.
- How Does an Agent System Grow from a Harness into an Ecosystem? — Lecture 9 of Nanjing University's Generative Software EngineeringStarting with the version-control challenges of parallel agents, this article examines reliability and changing requirements in agent products, then uses DeepSeek Harness's event bus and plugin model to explain how a thin kernel can support a developer ecosystem—and why plugins cannot remove semantic conflicts.
- How Does an Agent Move from One Generation to Long-Term Maintenance? — Lecture 8 of Nanjing University's Generative Software EngineeringNotes on Lecture 8 of the 2026 Nanjing University course Generative Software Engineering: starting from a minimal agent and the Ralph Loop, treating the workspace as development state, and covering CI/CD gates, parallel collaboration, and documentation-consistency practice in DeepSeek Harness, along with the limits of continuous cleanup, architecture refactoring, and chances to try and fail.
- Why Can Good Architecture Grow with Requirements? — Lecture 7 of Nanjing University's Generative Software EngineeringNotes on Lecture 7 of the 2026 Nanjing University course Generative Software Engineering: starting from structured programming and growable languages, understanding UNIX, relational databases, the Internet, MVC, and frontend state models, then using bookkeeping and the registrar system to show how saving facts and expressing dependencies turns a state consistency problem into a checkable computation.
- Why Do Requirements and Architecture Matter More Once AI Makes Code Cheap? — Lecture 6 of Nanjing University's Generative Software EngineeringNotes on Lecture 6 of the 2026 Nanjing University course Generative Software Engineering: AI generates code, tests, and prototypes quickly, while requirement conflicts, historical evidence, and the boundaries of change still have to be designed. The registrar system, the Towers of Hanoi, and the Excel QR code show that the core of architecture is finding a suitable representation for complexity and change.
- From Vague Requirements to Maintainable Systems: The Ins and Outs of Software Engineering — Lecture 5 of Nanjing University's Generative Software EngineeringNotes on Lecture 5 of the 2026 Nanjing University course Generative Software Engineering: starting from AI generating prototypes directly, a look back at the software crisis, Apollo software, Royce's critique, MVP, test-driven development, contracts, and UML, and at how AI makes implementation faster while leaving the gap between intention, specification, and implementation in place.
- Agents Push Development Down to Seconds, So How Does Project Management Stay in Control? — Lecture 4 of Nanjing University's Generative Software EngineeringNotes on Lecture 4 of the 2026 Nanjing University course Generative Software Engineering: once agents push development down to seconds, commits, conflicts, task breakdown, and traceability all need re-examining. The open-ended personal assistant assignment shows further that validating the core value first and extending the architecture afterward is more reliable than generating a complete system in one pass.
- Now That Agents Take Over the Operations, What Does Git Still Teach Us? — Lecture 3 of Nanjing University's Generative Software EngineeringNotes on Lecture 3 of the 2026 Nanjing University course Generative Software Engineering: how Git organizes software projects with traceable snapshots, why atomic commits, branches, builds, and tests still matter, and how people should hold on to architecture and engineering taste in the agent era.
- The Essence of Prompt Engineering: From a Single Instruction to a Full Context — Lecture 2 of Nanjing University's Generative Software EngineeringNotes on Lecture 2 of the 2026 Nanjing University course Generative Software Engineering: what an agent's context is made of, why a verifier matters more than role-play, and how to use context engineering for software development, paper reading, and long-horizon open tasks.
- Why Do We Still Need Software Engineering in the AI Era? — Lecture 1 of Nanjing University's Generative Software EngineeringNotes on Lecture 1 of the 2026 Nanjing University course Generative Software Engineering: the background of AI changing learning and work, how to use agents, and the software engineering method between intention, specification, and implementation.