Blogs / Ideas from the work
Thinking out loud.
07 essaysNotes on building AI systems, the decisions behind them, and the questions that deserve a closer look.
AI Engineering / System Design
What Building AI Agents Taught Me About Web Access and Permission
Lessons from building AI workflows at Yuba: how web retrieval differs from permission, where quality checks stop, and what explicit access boundaries require.
Applied Machine Learning / Aviation Operations
When Airline Crew Do Not Return: Forecasting the Staffing Risk
An engineering proposal for airline crew nonreturn: forecast staffing gaps, plan qualified reserves, and assess the limits of machine learning and LLMs.
Information Economics / AI Product Design
How AI Agents Could Change the Economics of Online Information
How AI agents could affect publisher revenue, why attribution is not a payment mechanism, and how to compare licensing models using explicit assumptions.
AI Engineering / Security
A Proposed AI Access Manifest for Web Content
A proposed AI access manifest for web content, with explicit rules for retrieval, reuse, retention, attribution, and decisions made outside the model.
AI Engineering / Evaluation
Testing AI Agents for Unauthorized Access and Data Use
A proposed framework for testing AI agent authorization, with observable failure criteria, synthetic Python checks, and clear limits on what results prove.
Agent Engineering / Privacy
Designing Task-Specific Privacy Permissions for AI Agents
A practical proposal for task-specific AI permissions covering data access, tool sharing, retention, revocation, and the limits of automated enforcement.
Agent Engineering / Product Design
Designing AI Agent Permissions Without Constant Approval Requests
A practical design for AI agent approvals that reuses scoped permissions, asks at meaningful boundaries, and blocks uncertain or unauthorized actions.