Clean Architecture and AI
summary
Claim: Clean Architecture matters more when LLMs generate code because layers and dependency rules stop agents from producing sprawling, unchecked changes. Recommendations include putting ArcUnit in CI to enforce boundaries, using Snyk and SonarQube for supply‑chain checks, keeping prompts small and specific, splitting roles across models and validating with generated diagrams and observability, and choosing OpenAI vs Anthropic tradeoffs to shape microservice and Kubernetes boundaries rather than leaving them to ad hoc agent commits.
AI-written summary · about
chapters (9)
- 0:00Introduction to talk
- 1:06AI for clean architecture
- 2:31Human-AI collaboration limits
- 3:52Code isolation and context
- 4:57Prompt specificity in AI
- 6:23CI pipeline validation
- 7:33Verifying AI-generated code
- 8:55AI code generation risks
- 10:40Test-first approach necessity
related talks
31mModern Architecture for AI-Augmented Development Teams
29mBuilding Trust in AI Agents
44mAspect-oriented Architecture
12mBeyond AI Hype: Architecting Software Systems That Scale in the Agentic Era
01h 01mSoftware Architecture for Gen AI
09mBest Practices to make your codebase AI ready
17mArchitecture panel
22mBuilding Better Software: A New Blueprint for the Agentic SDLC