Your AI Agent Parsed the Modbus Frame—Now Prove It Belongs to This Machine StateA practical state-ownership gate for stale frames, replayed events, recipe drift, and AI-assisted fluidics control.7 min read·fluidics-engineering-notes.hashnode.devAug 25
Your AI Agent Says the Pump Is Running. The Fluid Path May Disagree.A six-layer observability model for AI-assisted diaphragm-pump workflows: command, actuator, pressure/flow, bubbles, inventory, and route recovery.7 min read·fluidics-engineering-notes.hashnode.devAug 18
AI Agents Need Fluidic Observability Before They Can Run a Synthetic-Biology WorkflowA five-layer architecture for future AI agents to interact with real pumps, valves and sensors while deterministic controls enforce safety.5 min read·fluidics-engineering-notes.hashnode.devAug 18
Why Tube ID Tolerance Can Break a Fluidic Design: A Python Sensitivity CheckQuantify how tube inner-diameter tolerance changes laminar pressure drop using a dependency-free Python check and worst-case bounds.5 min read·fluidics-engineering-notes.hashnode.devAug 13