Reasoning with Receipts: Building Verifiable AI Systems
Exploring how deterministic reasoning and verifiable decisions can make AI systems more transparent and accountable.
Writing
Essays on building products, software development, and the intersection of technology and business.
The Context Paradox: When Knowledge Systems Contradict Themselves
How contradictory datapoints in knowledge bases create paradoxical instability based on which slice of context is visible at any given moment.
Symbolic Hallucination in AI Code Generation
When explanations impersonate execution, confident prose feels like progress while behavior stays the same.