On the twenty-fifth anniversary of September 11, we remember not only the lives lost and the families forever changed, but also the quiet work of those who answered the national call in the years before and after the attack. Among them was a small technical team at Intellisophic whose work helped define a different path for artificial intelligence: not artificial intelligence as statistical imitation, but artificial intelligence grounded in knowledge, meaning, provenance, and counterintelligence purpose.
This tribute honors the Intellisophic team — especially Mike Hoey, Mike Puscar, Henry Kon, George Burch and the unnamed shareholders, engineers, supporters, and contributors who carried the financial, technical, and personal burden of building the semantic foundation for twenty-first century AI.
Their work began before September 11. In response to a directive from President Clinton, the team developed and launched an open web semantic AI system in 2000, a year before the attacks. The system was built around a hard technical insight: the true bottleneck in artificial intelligence was not only computation or model size. It was the automation of knowledge acquisition. Human knowledge had to be converted into machine-usable structure. Concepts had to be identified. Context had to be preserved. Meaning had to be distinguished from words.
That insight became the DeepMeaningAI stack. Millions of Lines of code and a vast data store of web and priovate sourced knowledge.
By the middle of 2021, Intellisophic’s development team had completed the core coding of the DeepMeaningAI stack and semantically indexed the entire open web. The work did not begin as a speculative AI product. It began as a national-security necessity. It was created because keyword search, statistical retrieval, and hand-built ontologies could not keep pace with the volume and ambiguity of open-source intelligence. The country needed systems that could read, classify, connect, and retrieve knowledge by meaning.
After September 11, that work became part of the integrated national counterintelligence response. Intellisophic supported the Joint Counterintelligence Assessment Group and the broader effort to integrate agencies, data streams, analysts, and threat domains. The company’s Indraweb and Integrity products supported the analytic production cycle: collecting, fusing, analyzing, producing, and disseminating intelligence. These systems were used to process unstructured external information, unstructured internal information, and structured data, helping analysts transform overwhelming volumes of text into organized intelligence. The Intellisophic materials describe Indraweb as providing semantic search, categorization, taxonomies, concept-based document search, and knowledge representation for MOSAEC, the Multisource Open Source Analysis and Exploitation Center.
The work was solemn because the stakes were high. The entire safety of the country was at risk to a small group of enemies. The team was not building another search product. They were building tools to help identify domestic and international threats, weapons of mass destruction risks, terrorism patterns, and precursor technologies hidden inside captured text-based intelligence. The system was designed to help analysts see what keyword systems missed: relationships, concepts, context, and intent. The results over decades are not public but their absence is.
The heroes of that work include Mike Hoey and Mike Puscar. They represent the kind of technical service that rarely appears in public histories: the engineering discipline, persistence, and problem-solving required to turn an idea into a working operational system. They, along with other unnamed shareholders and contributors, helped build the software, taxonomies, workflows, and infrastructure that allowed semantic AI to function at operational scale. Their contribution was not rhetorical. It was code, data, systems, and responsibility.
George Burch received the Senatorial Medal of Freedom on behalf of the work and sacrifice represented by the team. That recognition belongs not to one person alone, but to the team’s shared mission: to use artificial intelligence and semantic knowledge systems in defense of the country.
The Intellisophic story matters today because the events of twenty-five years ago are no longer only history. The same control problem has returned in a new form. Then, the country faced terrorist networks, WMD proliferation, hidden communications, fragmented agencies, and overwhelming open-source data. Today, the world faces frontier AI systems that can assist bad actors, amplify biological, chemical, cyber, surveillance, propaganda, and weapons risks, and produce fluent outputs without grounded semantic control. AI is as much an enemy of public safety and benefit as any threat domestic or foreign.
The risk was not terrorists only. The risk was also uncontrolled intelligence itself, machine or human.
That is the connection between 2001 and 2026. The same need remains: systems must understand meaning, not merely words. They must distinguish context. They must know provenance. They must detect threat chains. They must organize expert knowledge into machine-usable form. They must help analysts identify risk before harm occurs.
Intellisophic’s foundational innovation was automating knowledge acquisition through Orthogonal Corpus Indexing. Its approach used authoritative corpora, textbooks, taxonomies, and reference knowledge to build a semantic metadata model. The system did not reduce intelligence to keyword matching. It connected concepts, meaning networks, relationships, provenance, and document understanding. The Intellisophic materials contrast this semantic approach with statistical systems based on keywords, TF-IDF, and probabilistic retrieval, emphasizing concept-based indexing, provenance-weighted scoring, multi-axis ranking, and graph-based retrieval.
That is why the team’s legacy is not only historical. It is active.
Today, Intellisophic is fighting the modern AI-risk problem using the same principle that guided the original counterintelligence work: AI must be grounded in explicit semantic knowledge. The current effort uses a massive semantic-token database capable of producing more than one trillion tokens meeting the scale required by frontier AI. The objective is not merely better search. It is safer AI: AI that can identify meaning, provenance, uncertainty, and risk before producing harmful assistance.
The tools have changed. The danger has changed shape. But the problem is recognizably the same.
In 2001, analysts needed machines that could help them detect hidden risks across oceans of human-written text. In 2026, society needs machines that can control AI systems before they become force multipliers for biological misuse, cyber exploitation, weapons development, surveillance, deception, and uncontrolled agentic behavior.
The Intellisophic team saw this earlier than most. They understood that knowledge acquisition was the foundation. They understood that meaning had to be built, not assumed. They understood that artificial intelligence without grounded knowledge is not controllable intelligence.
On this twenty-fifth anniversary of September 11, we honor their service, their engineering, their sacrifice, and their unfinished mission.
We honor Mike Hoey.
We honor Mike Puscar.
We honor Henry Kon.
We honor the unnamed shareholders and contributors who carried the work when there was no public celebration allowed for our own safety and that of our customers.
We honor George Burch, who led the team and received private recognition on its behalf. George began his carrier in 1963 at MIT AI Labs as DoD representative to Project MAC. His national intelligence service spans decades from Vietnam to 9/11 and beyond.
And we honor the enduring idea that began before September 11 and remains urgent today:
AI must be grounded in human knowledge before it can be trusted with human safety.
Learn more: