Technology & Intellectual Property

What if an AI could remember not just what happened, but what it believed?

What if your AI assistant knew which of its facts came from a signed contract and which came from a hallway rumor? What if a security system could tell what a request is really trying to do, not just what it says? What if two characters in a story could witness the same event, walk away believing different things, and change their minds only when the truth surfaced?

That is the kind of intelligence I build: systems that keep a real history of what they know, what they believe, how sure they are, and where each piece of knowledge came from.

My work rests on two technologies I created and own:

• AI Lattice Intelligence: the architecture that organizes persistent knowledge and coordinates the reasoning that draws on it.
• KnowledgeFormers: living knowledge structures that give each individual agent its own memories, beliefs, relationships, and goals.

Both run in working software today.

The Builder

Troy Lee Yohn

Independent AI Systems Architect · Technology Developer · Author

My work has always existed at the intersection of creativity, technology, and unconventional thinking.

Over the years, I’ve developed digital platforms, interactive experiences, and software systems driven by a desire to explore what technology could become rather than simply what it already does. That has meant building custom business software, designing loyalty and mobile-coupon programs for retail, and working for nearly a decade as a data architect. I’m self-taught; everything I build comes from figuring out how things actually work.

One project crystallized the idea behind everything I do now. At a Databricks conference, I built one of the event’s top applications. It encoded a database’s structure and business definitions into a knowledge graph, which let plain-language questions become precise queries and then real insights.

The lesson was simple. AI is most trustworthy when its knowledge lives in an explicit structure it can point to, not only inside a model’s weights.

AI Lattice Intelligence and KnowledgeFormers grew out of that conviction. They are my own independent work.

AI Lattice Intelligence

A smarter foundation for context

Most AI applications rebuild their context from scratch every time a request arrives. AI Lattice Intelligence gives a system a lasting, structured understanding instead, one that grows with every interaction.

Knowledge that lives outside the model. The language model is a component the system uses, not the place its knowledge is stored. Models can be swapped or upgraded without rebuilding what the system knows.

Provenance on everything. Every piece of knowledge can trace back to where it came from, how confident the system is in it, and whether it has been confirmed.

Understanding that changes over time. Conclusions are valid for a period rather than permanent. Knowledge that keeps being confirmed grows stronger, and knowledge that stops mattering fades.

Disagreement treated as signal. When evidence conflicts, the system weighs it from more than one reasoning perspective. Where they disagree, the disagreement is recorded rather than smoothed over.

Knowing what it doesn’t know. Gaps in knowledge are treated as information in their own right. They are not blanks to be filled with a confident guess.

Private by design. The architecture runs on locally hosted models and private infrastructure, so sensitive knowledge never needs to leave the owner’s control.

KnowledgeFormers

Every agent, its own point of view

KnowledgeFormers rest on one distinction: what is true is not the same as what an agent believes.

Each KnowledgeFormer holds one agent’s own memories, beliefs (with how strongly each is held), relationships, and goals. An agent can misread a situation, trust the wrong source, or hold a belief the evidence doesn’t support. Later, it can revise that belief when it learns something new. It never inherits an all-knowing view of the world.

A great storyteller needs exactly that. So does a trustworthy assistant, a realistic simulation, or any agent that should behave like it has actually lived through its experiences.

Working together

KnowledgeFormers are the knowledge of each individual agent. AI Lattice Intelligence is the architecture that connects that knowledge to reasoning, memory, and action. Together, they make intelligence that has a history.

Applications

One architecture, many fields

These technologies apply anywhere knowledge has a source, a history, and more than one point of view. Some applications are running today. Others are in active development.

Running today

Living story worlds. Nectara Viva, an audio-fiction platform from AI Fusion Foundry S.A.S., builds story worlds in which every main character is a KnowledgeFormer. An orchestration layer called the Fates creates situations and pressure instead of writing scripts. Each character responds according to its own knowledge and goals, within rules the creator sets.

Personal expertise systems. One system, built for a marketing consultant, combines established domain knowledge with her own proven methods. When she pushes back or proves an idea, her input updates the system’s foundations, not just a layer of personalization. It helps turn a professional’s experience into a framework they own.

In development

AI security and intent analysis. As AI agents act on our behalf, AI will increasingly be attacked by other AI. This work analyzes what a request is really trying to do before it reaches a model. That one capability can serve prompt-injection defense, agent-to-agent threat detection, and fraud detection.

Defense against manipulation. People face scams, high-pressure persuasion, and coordinated influence campaigns every day. A system that recognizes those tactics, and explains why something looks manipulative, helps people keep their decisions their own.

Private, loyal personal AI. This is an assistant that runs on infrastructure its owner controls, remembers what matters, and answers only to that owner. For executives and boards, it can mean strategic intelligence that never touches a public cloud. For individuals, it can track decisions made, ideas ruled out, and the reasoning behind both.

Enterprise knowledge with provenance. Most corporate AI search retrieves passages of text. A knowledge-centric approach models what an organization knows, where each piece came from, and how work actually gets done. When teams give conflicting direction, each path can be kept and attributed rather than forced into a single answer.

Natural-language data intelligence. Business data and the meaning behind it are mapped into a knowledge graph. People ask questions in their own words and get answers grounded in the actual data, with no need to know the database.

Connected business intelligence. Companies run separate AI tools for forecasting, inventory, fraud, and customer behavior, and those tools rarely talk to each other. A shared knowledge structure can model the relationships between them, such as across a supply chain, its financial flows, and the people who buy from it.

Retail and in-store experience. The same architecture can learn which displays, placements, and messages actually move shoppers, while keeping individual privacy built in from the start.

Knowledge-driven commerce. Obscure Find is a curated marketplace operated by Orange Umber LLC, with operational support from AI Fusion Foundry. It is exploring structured knowledge for discovery and search. It is a separate commercial application and does not own the underlying technology.

Interactive and embodied characters. Characters with persistent identity don’t have to stay on a page or in an audio series. The longer-term direction includes interactive companions, and eventually physical ones, that remember, respond in character, and grow with the people they know.

Personality for machines. What if every machine people live with had a real personality, one that remembers you, grows with you, and stays itself? Robots and assistants are becoming remarkably capable, yet each one meets every person the same way. KnowledgeFormers can give a machine its own identity, relationship memory, and character, which is what turns a device someone uses into one they love.

The opportunities include:

• Home and humanoid robots. The hardware is getting good. Personality and relationship memory are what make a robot part of the household.
• Companions for older adults. A companion that remembers your grandchildren’s names and your stories meets a real and growing need for connection.
• Toys. Toy makers are already moving to add AI, and a character that stays consistently itself matters most when the audience is young.
• Cars. An in-car companion with a consistent personality that knows its driver.
• Games and theme parks. Characters that remember each visitor and pick up where they left off.
• Brands. A mascot that can actually hold a conversation, with “never do” rules that protect the brand.

Intellectual Property & Licensing

Independently developed. Personally owned.

AI Lattice Intelligence and KnowledgeFormers were conceived and developed by Troy Lee Yohn. The underlying intellectual property is personally owned. That includes the original software, architectural designs, technical documentation, and proprietary implementations. It is held separately from affiliated operating companies.

Commercial rights are granted only through individually negotiated written licensing agreements. Ownership remains with the inventor.

Latin America

AI Fusion Foundry S.A.S. (Medellín, Colombia) is the primary commercialization partner for these technologies throughout Latin America, from Mexico through Central America, South America, and designated Caribbean markets. Affiliated companies, including Orange Umber LLC, may hold application-specific rights consistent with that regional position.

Licensing and collaboration

Opportunities may include:

• Technology licensing for AI products and applications
• Geographic or industry-specific rights
• Development of specialized agents and intelligent environments
• Research and commercial development partnerships

Technical details are shared only under a confidentiality agreement.

Intelligence Should Have a History

I believe the next real advances in AI won’t come only from bigger models. They will come from systems whose understanding persists, evolves, and can explain itself: systems that know the difference between evidence and belief, remember what matters, and grow wiser with experience.

That is what I’m building.

Technology and licensing inquiries: please reach out through the contact page.

© 2026 Troy Lee Yohn. All rights reserved. AI Lattice Intelligence and KnowledgeFormers are independently developed technologies, and all related intellectual property rights remain with their owner. Publication of these materials grants no license, assignment, or other authorization. Commercial use requires a separate written agreement.