zAfi: from guessing toward structured reasoning
A world where trusting AI means checking it, because every fact it uses has an address you can look up.
NanoSI is what we're building now. zAfi is where we think it leads.
The horizon
Our research bet is that more capable AI, and in the long run forms of superintelligence, should be built on structure that can be inspected and verified, so capability and accountability grow together. This is a long-term horizon, not a near-term claim.
Four research directions
Each of these is a hypothesis.
Tesseract-Lattice
Multi-dimensional knowledge representation. See the tesseract on Technology.
NSRM
Neuro-symbolic recursive models. A hypothesis, named here only.
NanoTRM
Small models that refine an answer over several passes. A hypothesis, named here only.
Thought Trace
A readable record of reasoning steps. A hypothesis, named here only.