Ideas Nobody Else Thought Of
These aren't concept papers or pitch decks. They're working systems, grounded in research. Each one started with a question that seemed off-tangent until you review the scientific literature.
The common thread: seeing connections that others miss, then building them.
aXread Assistant
AI that speaks your language
An AI assistant with Easy Read mode built in. One toggle transforms complex explanations into simple language with emojis, short sentences, and the option to listen instead of read. The people who could benefit most from accessible AI are often the last to get it—we closed that gap.
The insight: What if AI assistants were built for people with intellectual disabilities, not just screen reader users?
Savanax / Dialectica
Multi-perspective reasoning without compression
Most AI systems compress everything into averages. Savanax preserves disagreement. 19 reasoning branches analyse decisions through isolated corpora—the computational equivalent of savant cognition's reduced top-down inhibition.
The insight: What if we could suppress AI's Bayesian priors the way rTMS suppresses neural compression in savants?
AxelV1
Autonomy support for neurodivergent lives
Event-driven decision automation for complex support needs. Processes 24,000+ annual events—support worker messages, medication alerts, behavioural observations—through policy-based automation. Augments human judgment without replacing it. The person's expressed preferences are always the primary signal.
The insight: What if we could automate routine care decisions while preserving genuine autonomy and catching when support workers override the person's actual preferences?
Context Scanner
AI-friendly codebase documentation generator
Generates 9 structured context files that help AI assistants understand your codebase instantly. Scans architecture, APIs, data models, validation rules, and automatically detects compliance gaps against Australian accessibility (WCAG AAA), privacy (OAIC APP), security (ACSC Essential Eight), and AI governance (VAISS) frameworks.
The insight: What if your AI coding assistant could understand your entire codebase—architecture, patterns, compliance gaps—the moment it opened the project?
The Innovation Philosophy
Every innovation here started as a question that seemed tangential—until you dig into the research. "What if AI could reason like a savant?" "What if your nervous system could guide your wellness app?" "What if care automation could preserve autonomy?"
The difference between an unconventional idea and an innovation is implementation. I don't just wonder about these things—I build them and find out.
Got an idea that sounds too unconventional? Those are my favourite.