For Immediate Release
Bath, Ohio — May 4, 2026
Centralpoint keeps Skills and Prompts local, runs LLM-agnostic, and unifies data indexing and inferencing in a single platform — with the engagement model to prove it on customer data before any commitment.
Oxcyon today released the first on-premise AI Governance Platform: Centralpoint. Oxcyon is not a start-up. The company was founded in 2000 and has spent twenty-five years continuously developing the data-aggregation, transformation, governance, and inferencing foundation the platform is built on. Centralpoint is in production today across 85+ enterprise deployments — including the US Congress, FedEx, Samsung, Ericsson, RWJBarnabas, and Centene — serving approximately fourteen million governed user sessions per month. The on-premise AI Governance Platform announced today is the fifth era of that single architectural lineage, not a new entrant building toward enterprise readiness.
Watch the Announcement
In the announcement video, Oxcyon leadership walks through the heritage that produced the platform, the architectural principles that distinguish it from cloud-only AI vendors and from policy-only governance platforms, and the “install-and-prove” engagement model that lets prospective clients evaluate Centralpoint against their own data, in their own environment, before any procurement decision is made.
What Makes Centralpoint Different
Centralpoint is the first AI Governance Platform architected from the ground up to run on the client’s own infrastructure, behind the client’s perimeter, under the client’s identity and audit regime. The on-premise posture is not the only differentiator; it is the precondition that makes the rest of them possible.
-
Skills and Prompts kept local. The behavioral vocabulary the platform uses to govern AI behavior — encoded as Skills, assembled into Prompts — lives on the client’s premises, under the client’s ownership. Skills the user is not authorized to invoke are never exposed to the model in the first place, and proprietary instructions never leave the client’s boundary.
-
LLM-agnostic by architecture. Clients choose their generative-model providers (OpenAI, Anthropic, Gemini) and their embedded models (Llama, Qwen, Onyx). Substituting one model for another is a configuration change, not a re-implementation event. The deployment is never coupled to a single model vendor’s pricing surface, release cadence, or commercial posture.
-
Data indexing and inferencing unified in one platform. Centralpoint aggregates the client’s existing content estate in place — SharePoint, Office 365, Google Drive, OneDrive, JSON, XML, Excel, databases, email archives, scanned paper, legacy formats — transforms it through the governance and classification layers, builds the hybrid retrieval index, and runs the inference call against that index, all under one roof. The client does not assemble four vendors to produce one governed AI deployment.
-
On-premise deployment, with hybrid and cloud options available. The AI Index can stay on the client’s hardware while inference processing runs in the client’s private cloud tenant; the boundary is enforced by the platform regardless of which side each component sits on. Hybrid is a deployment configuration, not a separate product.
“Oxcyon has been developing the most robust on-prem aggregation, inferencing, and mining platform for over 25 years. We've agnostically integrating with all LLMs as the next logical step.” “We have further recognized the importance of any organization's SKILLS and PROMPTS as their institutional knowledge (capital) and believe shouldn't be readily shared. Those unique processes and methods are your ‘special sauce. Centralpoint empowers you to securely manage skills and prompts securely. This redefines traditional RAG, increasing performance, speed and relevance. This also allows metering against tokens or fees, reducing calls to any LLM beyond the second identical request made.’”
— Samuel Keller, CEO and President, Oxcyon
▸ Proof Before Purchase
We don’t demo, we deploy. Oxcyon installs Centralpoint in the prospective client’s environment, points it at the client’s actual data, runs the platform’s aggregation and transformation pipeline against the client’s actual content estate, configures the deployment under the client’s actual role and audience model, and lets the client exercise the deployment against the queries the deployment will be expected to handle — all before any procurement commitment. The evaluation surface is the production architecture. The proof is not a curated reference deployment; it is the platform, running on the client’s materials, in the client’s environment, before the contract is signed.
Read the White Paper
The full architectural treatment, comparison tables across five competitive cohorts, and a procurement objection-handling appendix.
Download the White Paper →
About Oxcyon
Oxcyon is the architect of Centralpoint, the AI Governance Platform now in production across 85+ enterprise deployments including the US Congress, FedEx, Samsung, Ericsson, RWJBarnabas, and Centene. Founded in 2000 and headquartered in Independence, Ohio, the company has spent twenty-five years building the data-aggregation, transformation, governance, and inferencing foundation that enterprise AI requires. Centralpoint is recognized in Gartner Magic Quadrant evaluations and serves as the operational platform behind approximately fourteen million governed user sessions per month. Learn more at www.oxcyon.com.
###