Ashtang AI: AI Architecture for Complex Ecosystems
summary
Claims a concrete eight-part AI architecture (Ashtang = eight parts) that layers digitization, interpretation (embeddings/knowledge graphs), causality, prediction, reasoning, controllability, simulation and adaptation; enumerates five types of causality and five reasons to build prediction models; uses knowledge graphs, APIs, embeddings and agent/workflow chains to turn telecom and retail examples (call-drop root causes, Ola demand, Spotify embeddings) into a state‑flow world model and digital twin simulation; argues LLMs are token prediction, reasoning requires explicit programmatic state transitions.
AI-written summary · about
chapters (11)
- 0:00Ashtang AI framework
- 2:32AI systems architecture
- 4:58Digitization layer foundation
- 6:24Data to meaning interpretation
- 9:32Causality for complex systems
- 14:03Prediction beyond forecasting
- 17:55Reasoning beyond language
- 22:12Controllability for insights
- 24:38Simulation for planning
- 26:43Adaptation for ecosystems
- 28:31State flow architecture
related talks
59mThe evolution in AI thinking and products of the next decade
29mWhat on earth does it mean to build AI software?
43mAI Systems: Foundations for Artificial Minds
55mAI is Coming - Now What? Architecting the Future of Software
47mReference Architecture of AI in the Cloud
31mAI, Digital and Value Streams Enable Data Driven Organizations
29mAgentic AI: What It Is and Why It Matters
01h 26mThinking in Graphs: The Cognitive Foundations of Agentic AI