The Intersection of Fitness Function driven Architecture and Agentic AI
summary
Claim: small, 10–15 line architectural fitness functions actually prevent big failures—examples include ArcUnit (Java), ArcUnit TS, Piest Arc and Arc Go catching cyclic dependencies; a log-based referential-integrity checker reads 24 hours of service logs to assert tickets map to customers. At enterprise scale brittleness appears, so projects expose an MCP/meta-bounded-context capability (service-side MCP adapter) that lets agentic AI invoke intent-level checks; metric normalized distance from the main sequence flags AI-generated “slop.” This approach works in practice to separate intent from implementation.
AI-written summary · about
chapters (9)
- 0:00Introduction to talk
- 0:52Architectural fitness functions
- 3:00Preventing cyclic dependencies
- 6:11Fitness functions for microservices
- 10:06Data integrity in distributed systems
- 12:58Enterprise architecture governance
- 15:06Agentic AI and MCP integration
- 19:04Fitness function-driven architecture
- 23:25Summary and future outlook
related talks
44mAspect-oriented Architecture
42mFitness Functions for Your Architecture – in Practice
01h 03mCreating and Maintaining Architectural Fitness Functions
11mClean Architecture and AI
36mEngineering Intent: Solving the AI Productivity Paradox
50mBuilding Evolutionary Architectures
17mArchitecture panel
58m4 Key Trends Towards Better Software Architecture