Chad Houghton

Director of Technical Product Management. Building AI-powered systems and methodologies for structured multi-model workflows.

About

Based in San Diego. I design and build systems that coordinate multiple AI models to produce better technical outcomes than any single model operating alone. My work focuses on the infrastructure between human intent and model output — prompts, directives, validators, and dispatch protocols.

Currently building LLM Crossfire, a multi-model deliberation tool, and maintaining a set of open-source frameworks that formalize how to work with frontier AI assistants.

Projects

LLM Directive Framework

github.com/choughton/llm-directive-framework

Six directives that fix the five worst AI assistant behaviors. Tested across Claude, Gemini, and ChatGPT with platform-specific deployment guides and documented findings. Includes testing scenarios and prompts to validate changed behavior.

Feature Design Templates

github.com/choughton/feature-design-prompt-templates

A 15-template adversarial pipeline for multi-model ideation, problem identification, feature design, and work breakdown. Three frontier LLMs review in parallel to surface anchoring, blind spots, and false convergence throughout the process.

LLM Handoff

github.com/choughton/llm-handoff

A file-based dispatch protocol for multi-CLI AI coding workflows. Prompts are advisory; validators are authoritative. Uses markdown files for managing state, backlog, and handoffs.

Contact