It did not start with ChatGPT. The people behind Intelligence Factory have been deploying AI in transportation, aviation, supply chain, telemedicine, and healthcare since 2009. Today, that work is built with Buffaly, the open-source neurosymbolic runtime we use every day.
It connects language, typed tools, semantic memory, and workflow execution so AI can participate in real operations instead of staying in a chat box.
Orlando has healthcare systems, billing operations, simulation teams, logistics companies, independent pharmacies, and founder-led businesses with practical work that AI can help move forward.
Intelligence Factory brings a long local history to that opportunity. Since 2009, our work has included customer engagement systems for AT&T and Delta, telemedicine for MEDEK out of Mount Dora, aviation ERP work on multi-terabyte Oracle databases, supply chain monitoring for Halliburton, and clinical operations for hundreds of independent practices through FairPath.
That mix of Orlando roots, healthcare experience, and Buffaly-based AI infrastructure is what makes this work fun for us. We get to build systems that are useful, inspectable, and strong enough for real operations.
Intelligence Factory is a technical AI company based in Orlando, Florida. We build AI systems for healthcare, billing, regulated workflows, and operational businesses where reliability, auditability, and data control matter.
Buffaly began as a research project on how language attaches to meaning. The core idea: a child does not learn language from text alone - they learn it from repeated contact with the world, where every word is tied to objects, actions, situations, and outcomes. Matt Furnari, our CTO, has been circling this problem since the late 1990s, working through evolutionary programming, rule-based systems, neural networks, reinforcement learning, and stranger ideas along the way. Buffaly is the accumulation of the ideas that survived contact with implementation.
We use Buffaly internally to run development, client operations, planning, and workflow execution. It is not just something we sell. It is the runtime we build with and trust for our own work.
The company is organized around named systems in production, not service categories. Each of these is real, deployed, and linkable.
Typed executable graph, semantic entities, ProtoScript working medium, native C# execution, runtime tool creation. Open source on GitHub. In one production telemetry sample: 1.2 million messages, 380,000 tool calls, and 70% of new tools used in the same session they were written.
Spun out of Intelligence Factory. Bootstrapped from zero to tens of thousands of patients, 10 million remote care minutes, and 8-figure revenue. Runs CCM, RPM, RTM, and APCM across 200+ practices and MSOs with 98% payment success on remote care claims.
A unified ontology layer covering Epic, eClinicalWorks, Athena, Kareo, Practice Fusion, and one-off installations coded by a doctor's nephew. Fix a mapping once and every downstream rule works without touching the source EMR again.
Ontology-guided semantic data layer built for legacy, fragmented, and regulated data. Used to organize 10,000+ unlabeled call recordings into a defensible, structured response inside a regulatory audit deadline.
Buffaly is where our ideas about language, memory, tools, and real workflow execution come together. It gives our Orlando team an inspectable runtime we can improve and adapt for real operations.
In healthcare, civil monetary penalties can reach tens of thousands of dollars per HIPAA violation. Buffaly keeps PHI behind typed runtime handles - the model can reason about a patient or a claim without ever seeing the underlying record.
Production AI has to select steps, call tools, handle exceptions, and produce a trail someone can review. After operating voice systems across millions of patient interactions, we learned that model-driven compliance is unworkable - boundaries between language generation and regulated operations are mandatory.
Most AI agents pay full token cost on every run. Buffaly promotes repeated patterns into native code. In one production FairPath task, that meant a 79.7% reduction in token cost - and the system gets cheaper the more it is used.
Buffaly is open source. The runtime your workflows run on is published, inspectable, and not controlled by any single vendor - including us. If you stop working with Intelligence Factory, your AI systems keep running.
These are public examples of the kind of work behind the Orlando AI page: healthcare operations, clinical data, remote care, prioritization, and Buffaly itself.
We publish and maintain technical material on neurosymbolic AI, healthcare workflows, agent infrastructure, operational automation, and the practical limits of current LLM systems.
Telemetry from a real Buffaly instance: 1.2 million messages, 380,000+ tool calls, and 70% of newly created tools used in the same session they were written.
Why leading LLMs scored under 9% accuracy on a 1,150-term clinical mapping benchmark, and how ontology-grounded validation fixed it.
The origin story: dual-channel learning, sememes and lexemes, and why Buffaly did not start as a reaction to ChatGPT.
Ecosystems become real when serious people start comparing notes. If you are building, researching, funding, buying, regulating, or operating AI in Central Florida, we want to know what you are working on.
Orlando has the ingredients for a serious AI ecosystem: healthcare, simulation, education, defense-adjacent technical work, logistics, hospitality operations, professional services, and founder-led companies that move quickly. The people behind Intelligence Factory have been part of that ecosystem since 2009 - it is time more of it compared notes.
Follow Matt Furnari on LinkedIn for Central Florida AI notes, healthcare AI work, and practical updates from the team.
Buffaly is the open-source runtime behind this work. Visit the Buffaly site to learn how the runtime works and where the project is going.
The workflow triage is for organizations that already know they want to do AI and need help picking the right first project. Five business days, fixed scope, three documents you can take to leadership.
Whether you are evaluating AI for your organization, building something technical, looking for a local AI community, or trying to figure out whether AI belongs in your operation at all, Intelligence Factory is where to start.