Each project narrowed—or reopened—the same interaction question.

01 / private origin

myDNA: chosen context moved into explanation.

The useful contribution was an interaction: inspect structured results, focus a relationship, then carry a chosen item into an explanation space. Medical claims and personal output are not part of the current lab.

02 / public graph · 2024

AWS Glossary Graph: a vocabulary became navigable.

Started and finished in 2024, the public graph paired AWS services with the terms that defined or connected them. Search, filtering, neighborhood focus, and relationship inspection gave the canvas a practical reason to exist.

It demonstrated open exploration, not a validated lesson. Its interface is evidence for the Learning Toys branch, not a visual template for method experiments.

03 / broad hypothesis · 2025

ILP 2025: several promising interactions, one oversized claim.

The first ILP combined scripted synthesis, a pathway challenge, a contextual guide, and a small graph under a general AI-platform story. The interactions were real; ingestion, generation, adaptation, and impact claims were not.

The four interaction arcs now return as transparent, deterministic Learning Toys, while this page keeps the original states visible beside them.

04 / focused product outcome

CertGraph: traceability became a complete learning loop.

CertGraph narrowed the work to dense AWS certification scenarios: exact phrase, normalized constraint, option impact, surviving decision, then a distinct architecture proof.

Its verified structure informed ILP’s authored evaluator and traceability contract. CertGraph is not an ILP runtime dependency, content source, or visual design system.