About Intelligence Factory
AI should be safer, more controlled, and more explainable.
Not just more powerful.
Intelligence Factory is a software and AI company founded by Matt Furnari and Justin Brochetti. We build systems for organizations that need AI to perform real work under real controls.
Our foundation is Buffaly, a neurosymbolic architecture that connects model reasoning to explicit knowledge, executable tools, and rules the model cannot simply talk its way around.
The conviction
Giving a model more intelligence does not, by itself, give a business control over what it may do. Asking an agent to follow a rule is not the same as enforcing that rule.
We are not waiting for a larger model to solve the problems of authority, memory, and accountability. We are building those responsibilities into the architecture.
That distinction is why we built a different foundation. And why we built a company around putting it to work.
What Intelligence Factory does
We develop the architecture, build the software around it, and put it into operating use. The research and the delivery belong together.
An open-source neurosymbolic platform developed by Matt Furnari and commercially supported by Intelligence Factory. Model reasoning connects to explicit knowledge, typed tools, semantic memory, and executable workflows.
Explore the architectureCompany-specific systems and workflows. We connect applications, APIs, databases, and operating knowledge, then build what is missing. The objective is a working operation, not another interface that leaves people to finish the process manually.
Deployment and integrationOur remote-care operating platform for practices and pharmacies running APCM, RPM, CCM, and RTM. Patient work, requirements, exceptions, and billing readiness share an operating workflow. Clinical decisions remain with authorized practice staff.
See FairPathWhat makes Intelligence Factory different
Our difference starts with what we refuse to leave entirely to a model: authority, memory, execution, and the record of what happened. Those responsibilities need explicit structure.
In our published controlled-document demonstration, premature approval, duplicate submission, and author self-approval were blocked and recorded. An authorized reviewer's valid approval succeeded.
The compiled action enforced the rules, regardless of what the agent requested. Hiding an invalid tool from a menu is not the security boundary. The action itself checks the rules.
Read the demonstrationCONTROLLED DOCUMENT / PUBLISHED TRACE
Every attempt recorded. Invalid state changes prevented.
Knowledge, relationships, tools, and procedures become explicit objects the system can inspect and use. The record shows what was attempted, what was blocked, and what actually changed, not merely what the assistant said happened.
Buffaly can keep a native object inside the runtime while giving the model a typed reference to it. Tools operate on that object without repeatedly placing the underlying record into model context.
A procedure should not need to be rediscovered every time it runs. Buffaly turns successful work into persistent entities, actions, and code that can be reused and improved.
Read the learning and reuse evidence ↗Evidence, not a promise
In our published MedQA comparison, GLM 5.2 answered 101 of 150 questions correctly alone and 144 with Buffaly supplying enforced SNOMED CT retrieval and structured context. The model weights did not change.
Read the comparison ↗Our published learning analysis tracked 180 critic-created action/source-session pairs receiving later use across 89 working sessions. The record includes retained code, reuse, corrections, and outcomes, not a claim that every call succeeded.
Read the operating evidence ↗Who uses Intelligence Factory
We work on operations where information, people, systems, and exceptions have to come together. Our published work spans healthcare, customer engagement, aviation, supply-chain monitoring, and other complex operating environments.
If the requirement ends at generating text, much of this architecture is unnecessary. Its value becomes clear when AI must act, preserve state, respect authority, and leave a reviewable record.
Our published customer stories include TUMI Medical Corporation, whose founder, Dr. Jose Agusti, describes the importance of clear, auditable clinical and billing decisions.
The team behind Intelligence Factory
Matt Furnari and Justin Brochetti founded Intelligence Factory around complementary technical and business experience. Buffaly's research history predates the current wave of LLM agents.
CO-FOUNDER / CTO
Matt's research into language, learning, and representation reaches back to the late 1990s. Watching his young son connect words in children's books with pictures, actions, and experience helped sharpen the question that still shapes Buffaly: how can a system connect language to meaning it can actually use?
“I was not trying to build a system that merely produced plausible language. I was trying to build a system where knowledge could be represented, transformed, executed, and inspected.”Read the road to Buffaly ↗
CO-FOUNDER / CEO
Justin brings company-building experience and a business perspective to the partnership. In his published account of building three companies, he describes why technical capability needs a complementary understanding of customers, sales, operations, and business strategy.
A different architecture has to become something an organization can adopt, operate, and put to productive use. That is where the technical and business partnership matters.
Read Justin's perspective ↗How working with Intelligence Factory works
We start with the result you need and the constraint preventing it. The architecture serves that operating problem, not the other way around.
What needs to improve, and what is getting in the way?
Establish the systems, information, people, and exceptions.
Identify integrations, data boundaries, permitted actions, and approval points.
Connect the platform, tools, services, and domain knowledge.
Establish appropriate monitoring, maintenance, recovery, and support.
Key facts
| Company | Intelligence Factory, LLC |
|---|---|
| Founders | Matt Furnari and Justin Brochetti |
| Leadership | Matt Furnari, CTO Justin Brochetti, CEO |
| Roots | Orlando, Florida |
| Platform | Buffaly |
| Product | FairPath |
| Website | intelligencefactory.ai |
These details are the canonical description of Intelligence Factory.
Frequently asked questions
We do not ask the model to be the entire system. Buffaly connects model reasoning to explicit knowledge, executable capabilities, and runtime controls. Important parts of the work become inspectable and enforceable outside the conversation.
We mean defining and enforcing what an agent may do, rather than relying only on instructions asking it to behave. Our guarded-workflow demonstration shows invalid state changes being blocked and recorded even when the agent is told to attempt them. Authority remains in the guarded action.
Yes. Buffaly uses models for language, interpretation, code, and flexible reasoning. The architecture does not make a model solely responsible for memory, permissions, execution, or operating state.
We publish architectural explanations, executable workflow demonstrations, model comparisons, and analyses of real operating history. FairPath provides a healthcare operating context alongside that research. Each result is presented with its own scope.
Yes. Intelligence Factory connects APIs, databases, internal applications, and existing .NET software to Buffaly. Missing capabilities can be added as typed actions, services, transformations, and domain knowledge.
Intelligence Factory is the company. Buffaly is the open-source neurosymbolic platform developed by Matt Furnari and commercially supported by Intelligence Factory. FairPath is an Intelligence Factory product applying this approach to remote-care operations.
Start here
Tell us what the work requires. We will start with the operation, the controls it needs, and the result you want to achieve.
Talk to Us