Investor overview
The checkpoint that decides whether an AI’s information is good enough to act on.
Before an AI system acts, TruVector combines reader interpretations, subject guards, source dependence, and a separate instruction-consistency check. It returns Allow, Review or Block: an answer a machine can follow and a person can check.
Thesis
The enterprise risk has moved, and the tooling has not.
While AI systems only wrote text, a person read the output before anything happened. Now those systems file records, move money, change settings and call other systems on their own. The risk becomes operational and legal, and the human review step is exactly what was removed to make the automation worth it.
The question stops being whether the AI said something wrong and becomes whether it acted on something wrong.
The research opportunity is to evaluate an explicit checkpoint that combines semantic interpretation and source-aware aggregation before an action. Controlled comparisons establish its value relative to majority readings, deduplication, and specialized entailment methods.
TruVector is designed for integration into retrieval and agent workflows, with a reconstructable record of the readings, weights, thresholds, and action assessment behind each decision.
Research basis and evaluation
The reported preliminary 60-statement comparison produced 58 correct decisions for the reading-based replacement versus 14 for the earlier gate under the original labels. Two labels require independent wording review. The comparison is statement-only and does not estimate deployment accuracy. The whitepapers define the policy, study scope, and independent evaluation criteria.
Founder
Measurement is the discipline, not a metaphor.
TruVector is built and run by Michael Brandon Lane, from Johnson City, Tennessee.
His training is in chemical engineering and chemistry, where a number is worth nothing until you can say how it was measured and within what error. That is the whole idea behind this product, applied to information instead of materials.
- B.S. Chemical Engineering, Virginia Tech
- M.S. Chemistry, University of North Carolina at Greensboro
- Named inventor, U.S. Patent 11,949,124 (WO 2019/143483), coated lithium-ion battery separators
He designed and built TruVector and every other InTellMe product himself. The science behind it, Lane Vector, studies signal direction, temporal change, and source dependence through defined mathematical models.
That means a short line between a decision and a shipped change, from a founder who has already taken a product from the lab into high-volume manufacturing.
- Sibling site — intellmeai.com
- Direct — brandon@intellmeai.com
Investment
Self-funded so far. Raising a pre-seed round.
InTellMe has been built without outside money. The round will fund TruVector research and development, the growth of the products built on the same work, and company operations.
Terms, structure and financials are in the investor materials.
Request
Request investor materials.
The technical brief, test protocol, roadmap, use of funds and IP position are sent after review.