Common questions about Cognetryx

Cognetryx is the company behind Lumen, a private AI platform built for regulated enterprises. Lumen deploys inside the institution's own network, indexes the institution's own documents and policies, and runs the AI model privately. Every answer Lumen gives points back to the exact line in your own files it came from, so any answer can be checked. Cognetryx serves customers in healthcare, financial services, legal, government and defense, and manufacturing, where the question is not whether AI is useful but whether the AI's architecture can satisfy compliance and audit. Founded by Keith Kennedy, CISSP, with nearly two decades of regulated IT and security architecture experience.

Compare Lumen with building your own.

Lumen is the private AI platform from Cognetryx. It runs on your own servers or in a cloud account your team controls, answers questions from your own documents with a citation for every claim, and includes agents, meeting capture with Ekho, Word and PowerPoint output, and a Compliance Portal that logs every chat and agent run. Lumen works with the AI model your firm chooses.

Most enterprise AI runs in a third-party cloud, which means regulated data has to leave the institution's network to get an answer. For a hospital, a bank, a law firm, or a defense contractor, that is the compliance problem hiding inside every AI pitch. Lumen solves it architecturally: the model runs inside the institution's own environment, the documents stay inside, and the audit log of every query and citation stays with the institution's other regulated records. The deployment choice answers the questions an examiner will actually ask before the institution writes a policy or a validation report.

See how it works.

Lumen is built to support the architectural requirements of the major regulations that govern data in U.S. regulated industries. These include: HIPAA (45 CFR 160 to 164) for healthcare. The Gramm-Leach-Bliley Act and the Federal Reserve, OCC, and FDIC revised model-risk guidance (SR 26-2, April 2026) for banks. FDA QMSR / 21 CFR Part 820 and ISO 13485:2016 for medical-device manufacturers. NIST SP 800-171 Rev 2 (110 controls) and CMMC Level 2 for defense contractors handling Controlled Unclassified Information. ABA Model Rules 1.6 and 1.7 on confidentiality and conflicts for law firms. Cognetryx does not certify a customer's compliance program. The architecture removes the data-residency and vendor-handoff problems that block compliance in cloud-only AI deployments.

The longer view on each industry: healthcare, banking, legal, government, manufacturing.

It does not leave your network. Lumen deploys inside your environment, either on-premises or in a private cloud your team operates, indexes your documents privately, and serves the model on your infrastructure. Prompts, answers, the source passages the model cited, and the full audit trail all sit inside the boundary your security team already protects. There is no outbound call to a third-party AI service to get an answer, which removes a category of breach exposure, vendor-oversight obligation, and data-residency risk that cloud AI cannot remove. This is the architectural difference behind every other answer on this page.

Read the security model.

Lumen uses a fixed-cost subscription model rather than per-token cloud billing. Once the platform is deployed, query volume does not add to the price. That makes total cost predictable for institutions running AI across many users and workflows, where cloud-AI per-token pricing tends to compound faster than budgets anticipate. Pricing scales with deployment scope and your environment. Specific pricing is provided after a short scoping conversation, because the right number depends on your data shape, user count, and infrastructure.

Book a demo to start that conversation.

Request a live demo. In 30 minutes we ask about your setup, show Lumen answering questions from real regulated content with citations, and tell you whether it fits, along with honest answers on timing and hardware. Firms still scoping the problem can bring those questions to the same call.

Private AI is AI that runs inside your own network. The model, the documents it reads, and the record of every answer stay on servers you control. When an examiner asks where the data went, the answer is your own servers.

On-premises AI is one kind of private AI. It runs on hardware in your own data center. Private AI also covers a cloud account your own team runs, such as an isolated AWS or Azure network. Either way, the model, your documents, and the logs stay under your control.

An air-gapped LLM is a language model that runs on a network with no link to the internet. Prompts, answers, and documents stay inside that closed network. Firms use it when policy bars any outside connection, as in classified defense work.

No. Lumen deploys inside your own network, a private cloud you operate, or an isolated AWS or Azure VPC. The model, your documents, and the audit log all stay inside your environment, and the platform uses no external APIs for core AI processing.

Lumen is model-agnostic. Your firm chooses which AI model it runs, and your team decides when that model changes. The model runs inside your environment, next to your data, and Lumen connects to your systems through their APIs and to the files on your own servers.

We reply to demo requests within 24 hours and run a 30-minute demo on real regulated content. From there, the next step is a technical scoping call with your IT and security team, followed by a 90-day pilot inside your environment.

Cognetryx was founded in 2026 by Keith Kennedy, CISSP, and co-founded by Brent Fisher, and is based in New York. Its advisory board includes Julia Cherashore, former Deputy Superintendent for Data Governance and Management at the New York State Department of Financial Services.

Yes. Lumen is built to the standards regulated industries require, but any privacy-forward company that wants to use AI on its own data without handing that data to a cloud vendor is a fit.