Verifiable AI for high-stakes work

The idea in one minute

Every fact gets a four-part address, so every answer can be traced back to its source.

A square locates things in two directions. A cube adds a third. A tesseract adds a fourth. You can’t see it directly, but you can see its shadow, the rotating shape above.

Most AI keeps what it “knows” blurred inside billions of numbers. You can’t point to where a fact lives or where it came from. NanoSI does the opposite. It places every fact at a precise point in a four-dimensional lattice. One direction says what it’s about, one says how it connects, one says where it came from, and one says when it was true. Because each fact has an address, every answer can be walked back, step by step, to its source.

The tesseract, explained

Why now

• Small open models became capable. See the SmolLM3 model card.
• Fine-tuning got cheap. See the PEFT quantization guide (QLoRA) and Dettmers et al., QLoRA.
• Tooling for running models on machines you control is published and documented, for example Ollama and vLLM.

What we believe is defensible

These are hypotheses, not claims of a granted right.

• The lattice representation.
• The curriculum method.
• The verification loop.
• The evaluation suite.
• The small deployment footprint.

Business model and go-to-market (planned)

The planned wedge is legal and compliance, then finance and risk, then enterprise knowledge.

Risks we're watching

• Graph-building cost.
• Coverage gaps.
• Retrieval plus frontier models as the competing baseline.
• Evaluation leakage.