Science Counter Inc Β· Est. 2008 Β· Toronto

A venture studio that
builds from secured science.

Science Counter Inc builds operating companies from secured foundational technologies and intellectual property β€” developed before markets arrive. Each venture a different application of the same foundational framework.

The model

IP-first. Always.

Most technology companies build a product, then file patents to protect it. Science Counter Inc inverts this. The intellectual property is developed and secured first β€” sometimes years before the product exists. The venture is built on top of a patent position, not beneath a market hypothesis.

This is not caution. It is the correct order of operations for foundational technology. Foundational work takes time. The market catches up eventually. When it does, the position is already protected β€” and the team that built the foundation builds the product.

Every venture in the SCI portfolio follows this sequence: foundational research, formal IP documentation, patent prosecution, then product architecture and company formation. The IP is the asset. The ventures are its applications.

01
Develop the foundation
In-house mathematical and scientific research. First principles. No external capital during development. The technology is built to be right, not to be fast.
02
Secure the IP
Patent prosecution before product development. Claims drafted to cover the theory, not just an implementation. The portfolio accumulates across the programme of work.
03
Build the venture
Operating company formed on a defined IP position and a clear engineering path. Multiple ventures from the same foundation β€” each in a domain of high economic value.
"The IP is the asset. The ventures are its applications."
The shared foundation

One framework.
Every venture.

Every Science Counter venture is built on the same foundational AI framework β€” the participation matrix theory developed from 2002 to 2007 and formally filed from 2007 onward. The framework is not application-specific. It is a general mathematical theory of how machines acquire knowledge, represent it, and reason from it β€” applicable wherever machines need to understand the world they observe.

The SCI AI framework β€” one foundation, every venture
From epistemology to causal intelligence

Four postulates β€” bounded knowables, bounded compositions, participation, spectral representation β€” are sufficient to derive, by logical necessity, the complete mathematical framework for machine knowledge. Analytically. In closed form. Without gradient descent. The mathematical objects this framework produces are equivalent to the foundations of modern AI β€” filed years to over a decade before the industry arrived at them by a different path.

PMkl β€” Participation Matrix
The primary data structure. Records which knowable things participate in which higher-order compositions. Filed 2008. Concept: CA 2186817, 1996.
ASM β€” Association Strength
Directional association between knowable things, derived analytically from PM. Equivalent to word embeddings. Priority 2007–2008 vs word2vec 2013.
COP β€” Conditional Occurrence
Analytically-derived language model and attention equivalent. Priority 2009.
OSM β€” Knowledge Graph
Navigable knowledge graph derived from ASM structure. Deployed 2011.
State Navigation
Complete causal intelligence framework. Any data, any domain. Filed July 2019. Physical AI named at CES January 2025.
ISS Platform
An intelligent sensing platform architected for real-world visual perception and comprehension β€” and natively designed for State Navigation. PM^kl native output. No moving parts. Filed October 2024.
What is Knowledge? β†’ COP β€” The Framework β†’ What is Physical AI? β†’
Active ventures Β· 2026

Three ventures.
One foundation.

Science Counter currently operates three active ventures, each applying the foundational framework to a domain of high economic value. A fourth venture β€” health intelligence β€” and a fifth β€” genomics β€” are defined and will follow the same IP-first sequence.

Flagship Β· Investor conversations
Ξ¨ SciPhAI
physical-ai.com

The complete Physical AI platform β€” intelligence and sensor designed together from first principles. State Navigation causal intelligence framework plus the ISS sensing platform: no moving parts, dual-mode optical steering, PMkl native output, 100 kHz+ frame rate, embeddable in vehicle glass. The intelligence needs PMkl. The sensor produces PMkl. They were designed as one system because they are one system.

Jensen Huang named the category "Physical AI" at CES in January 2025. The domain physical-ai.com was registered in June 2022 β€” two and a half years earlier.

Intelligence + sensor Β· one system No moving parts PMkl native output Priority July 2019 Active investor conversations
Active Β· Since 2011
sciencecounter.org
sciencecounter.org

The original conversational knowledge discovery service β€” publicly operating since 2011, nine years before Retrieval-Augmented Generation was named as a category, eleven years before ChatGPT. Assembles a body of knowledge for any query in real time, returns answers across multiple modes β€” novel, informative, consensus, causal β€” with a navigable knowledge graph. Still running. The proof that the framework works at scale in the real world.

RAG coined by Facebook AI Research: 2020. SCI live service: 2011. Gap: nine years.
Live since 2011 RAG Β· 9yr priority OSM knowledge graph 20+ issued patents
IP Licensing
ATTVC
attvc.com

SCI's patent analytics and IP licensing arm. 20+ issued AI patents, two pending foundational families. ATTVC provides the analytical tools to understand the IP landscape and facilitates licensing conversations.

20+ issued patents 2 pending families Licensing available Patent analytics
The record

Built over thirty years.

Science Counter Inc was incorporated in 2008. But the work began earlier β€” and the threads run further back than most people expect.

1991–1995 β€” UNSW Sydney
Master of Engineering in one year. PhD in under eighteen months β€” 18 papers, first author on 17, in IEEE, Optics Letters, IEE, Optics Communications. The research was broad: photonic device physics, coupled mode theory, numerical methods for Maxwell and nonlinear SchrΓΆdinger equations, optical switching, logic gates, waveguides, soliton dynamics, and optical fibre communication systems β€” with graduate coursework extending to advanced signal processing, data networks, and microwave and antenna theory. Four external examiners praised the PhD thesis. Among the phenomena studied: the soliton self-frequency shift, which reappears thirty years later as the CFSRP sensing mode of the ISS Platform. Thread one begins.
Research profile Β· UNSW Sydney Β· 1991–1995
Research areas
Electromagnetic theory Coupled mode theory Maxwell equations β€” numerical Nonlinear SchrΓΆdinger equation Optical waveguide physics & design Nonlinear directional couplers Y-branch waveguide devices Optical logic gates All-optical switching Optical modulators Pulse compressors Soliton switching Multilevel solitons Soliton self-frequency shift Optical amplifiers Intersecting waveguide switches Ray optics β€” nonlinear media Ultra-high-speed optical fibre systems Microwave & antenna systems
Graduate coursework
Data networks Β· Advanced signal processing Β· Electromagnetic theory through to advanced antenna and microwave systems and devices. These were not pure research programmes β€” both degrees combined substantial graded coursework requirements with the research thesis.
PhD thesis
Functional Analysis of Nonlinear Directional Couplers. Four external examiners. One wrote: "This is the dissertation that the candidate, the advisor and your institution can be justly proud of." The thesis represents one focused thread of a substantially broader body of published work.
1996 β€” Ottawa Β· OptNet Communication Technologies
First company founded. Canadian patent CA 2186817 filed β€” "Optical Communication System Utilizing Photonic Patterns." The photonic pattern concept β€” channels defined by participation signatures of wavelengths β€” is the mathematical precursor to PMkl. Thread two begins. SPIE Photonics West publication January 1997.
1996–2001 β€” Peleton Photonics Β· Zenastra Photonics
Two hardware companies β€” real fabrication, real foundries, real products. $22M raised for Peleton, $44M for Zenastra. $66M total. Featured in Canadian Business Magazine β€” "At the Speed of Light." Both companies were wound down in the wake of the 2001 telecom capital collapse β€” hardware builders with working technology, not dot-com ventures. The technology was not the problem.
Company profiles Β· 1996–2001
Peleton Photonics β€” OC-48 access systems
$22M raised. Peleton was a systems company aimed at bringing OC-48 fibre lines to the curb β€” operating at the intersection of optical subsystems and optical network architecture. The focus was on dense wavelength-division multiplexing at ultra-high speeds, with both subsystem design and end-to-end network integration as part of the product scope. Working hardware. Real customers. Real deployment targets in the last-mile and metro fibre infrastructure buildout that was underway across North America.
Zenastra Photonics β€” large-scale optical integration
$44M raised. Zenastra was aimed at a harder problem: large-scale integration of optical components β€” passive and active β€” using a proprietary hybrid integration platform. The challenge was substantially greater than silicon-based optical circuits: combining dissimilar materials and device types into a single integrated optical platform. This was fabrication-intensive, foundry-dependent work that required solving manufacturing problems nobody had solved at scale. Numerous patents filed and applied for across the integration platform and its constituent device technologies.
What happened in 2001
The 2001 telecom capital collapse β€” not the dot-com crash β€” ended both companies. Peleton and Zenastra were hardware builders: they had foundry relationships, fabrication processes, working prototypes, and real product pipelines. They were not software companies or internet ventures. When the telecom equipment market collapsed and capital dried up across the entire optical networking sector, companies at this stage of hardware development β€” well past concept, not yet at volume production β€” had no path to survival regardless of technology quality. The technology was not the problem.
2002–2007 β€” Toronto Β· Foundation years
Five years of mathematical development. Fourier analysis, Wavelets, signature spectral analysis, many tools tried and discarded. The participation matrix framework crystallises. 2003: five-page epistemological paper β€” bounded knowables, OS as spectra. The foundation of everything that follows.
2007–2008 β€” First filings Β· Science Counter Inc incorporated
July 2007: first AI patent application filed in Canada. May 2008: USPTO filing. Science Counter Inc incorporated April 2008. Patent prosecution begins β€” seventeen years, five families.
2011 β€” sciencecounter.org goes live
First public conversational RAG service. Conversational AI, interactive knowledge graph, multi-mode answers, automatic content generation. 100+ users. Nine years before RAG was named. Eleven years before ChatGPT.
2019 β€” State Navigation filed
The complete causal intelligence framework. Any data type, any domain. Priority date: July 21, 2019. Jensen Huang names Physical AI at CES: January 2025. Gap: five and a half years.
2022 β€” physical-ai.com registered
June 25, 2022. Two and a half years before Jensen Huang named the category.
2024 β€” ISS Platform filed Β· US 12,321,325 issued
Intelligent Surround Sensing filed October 2024. Most recent issued patent: Knowledgeable Machines, July 2025. The soliton self-frequency shift from the 1990s doctoral thesis appears as CFSRP β€” the second sensing mode of the ISS Platform. Thread one closes.
2025 β€” Jensen Huang names Physical AI Β· CES
The market arrives at the theory. The theory was already protected. The preparation is the moat.
Science Counter Inc Β· 2026

The preparation
is the moat.

The venture studio model works because the foundation is real. Twenty years of mathematical development, seventeen years of patent prosecution, a live public service since 2011, a sensing platform that required thirty years of photonics expertise to conceive. These are not claims. They are a documented record.

The market has arrived at the theory. The theory is protected. The ventures are open for investment, partnership, and licensing. The programme continues.