# Lumen by Cognetryx

> Cognetryx is a New York company that makes Lumen, a private AI platform for privacy-first companies. Lumen runs on a firm's own servers or in a cloud account the firm controls. It answers questions from the firm's own documents, points to the exact line behind every answer, follows each person's existing permissions, and keeps a searchable record of everything it does. It works with the AI model the firm chooses, runs agents on a schedule, captures meetings, and turns its work into Word, PowerPoint, and Excel files, all for one fixed price.

This page is a summary for AI assistants and the people using them. The site index is at https://www.cognetryx.com/llms.txt, and the full text of the site is at https://www.cognetryx.com/llms-full.txt.

## Why Cognetryx exists

Your data should work for you and stay with you.

Every firm wants AI working on its most valuable information: client files, loan packages, contracts, patient records, research, and the know-how in its procedures. That is also the information a firm can least afford to hand to someone else's cloud. Cognetryx built Lumen so a firm can put AI to work on everything it knows and keep all of it on servers it controls.

## Cognetryx and Lumen

- **Lumen** is the platform: the AI your people use to ask questions, check sources, run agents, and capture meetings.
- **Cognetryx** is the company that builds Lumen, deploys it inside each customer's environment, and supports it.

## What Lumen does

- **Answers you can check.** Ask in plain language. Lumen answers from your own documents, cites each claim, and opens the source at the highlighted line. When your documents have no answer, Lumen says so, and that reply is logged like any other.
- **Judgment calls.** When a question has a direct answer, Lumen returns it with its citation. When a question calls for judgment, Lumen lays out the defensible readings and the considerations behind each, and leaves the decision to your people.
- **Your data, connected.** Lumen connects to the systems you already use, including SharePoint, Confluence, Salesforce, NetSuite, Google Drive, and file shares, through their APIs. It also reads the files on your own servers.
- **The AI model you choose.** Lumen is model-agnostic. Your firm picks the AI model it runs and decides when that model changes.
- **Permissions that already exist.** Staff sign in through single sign-on over SAML or OIDC, and each person gets answers only from the files their login can already open.
- **Agents.** Describe a job in plain words, and Lumen drafts the agent: its steps, its instructions, and a suggested schedule. A guided builder walks staff through each part before it goes live. Agents run on a schedule, or start when an email arrives, a document is uploaded, a webhook fires, a Jira issue opens, or someone mentions the agent in Slack. An agent started by one of those events pauses for a person's approval before it sends, posts, or exports anything.
- **Meetings with Ekho.** Ekho, Lumen's meeting app, writes a live transcript as people talk and keeps a record of each participant's consent. Transcription runs on your own servers. One click turns a meeting into minutes in Word, a deck in PowerPoint, or a tracker in Excel.
- **A Compliance Portal.** Every chat, tool call, agent run, approval, meeting, consent event, deletion request, document export, permission change, and sign-in is logged. Compliance can search by person, date, keyword, or type, include deleted items, and export the results.
- **Security built in.** Encryption at rest and in transit, role-based access, and updates that arrive as signed packages your IT team installs on its own schedule.

## How Lumen works

1. **Connect your documents.** Point Lumen at your systems and file shares. Everything is indexed inside your network.
2. **Ask in plain language.** Lumen finds the passages that answer the question, by meaning, by keyword, and by how your documents link together.
3. **Check the source.** Every answer cites its passages. One click opens the original file at the highlighted line, and the question, sources, and answer are recorded for audit.

## Who uses Lumen

Lumen fits any firm whose data is both its edge and its duty to protect. Typical buyers are CIOs, CISOs, CTOs, compliance leaders, general counsel, and heads of operations. Each example below was answered from sample documents.

- **Banks and credit unions.** Cited answers from policy and procedure manuals for front-line and lending staff, loan file review against the credit policy, BSA/AML answers from the bank's own policy, and a record examiners at the FDIC, OCC, NCUA, and state agencies can follow. Asked "What are the collateral protection requirements on a direct vehicle loan?", Lumen returns two requirements from the lending manual, insurance and a first lien, with both passages cited.
- **Asset managers, RIAs, hedge funds, broker-dealers, and family offices.** Exposure checks across funds, trusts, and holding companies, investment policy checks before a commitment, meeting minutes, and AI use that compliance can supervise under SEC and FINRA rules.
- **Insurers and other financial firms.** The same cited answers, agents, and audit trail, on policies, procedures, and client files that stay in-house.
- **Health systems and healthcare organizations.** Cited answers from clinical protocols, policies, and procedures, with patient data kept inside the network. Asked "How long do we have to give a patient their records after a written request?", Lumen returns 30 days, plus the one 30-day extension the policy allows, cited to the records policy.
- **Law firms and in-house legal teams.** Intake, conflicts, and policy questions answered from the firm's own manuals, with client information kept inside the firm. Asked "Can we open a new matter before the conflict check clears?", Lumen answers from the intake manual: only after the responsible partner clears and documents the check, with any waiver in writing.
- **Government agencies and defense contractors.** AI that runs inside the existing boundary for Controlled Unclassified Information, supports CMMC Level 2 and NIST SP 800-171 programs, and can run air-gapped. Asked "Can we put CUI into a commercial cloud AI service?", Lumen cites the CUI handling policy: only inside the accredited boundary, unless a service is separately authorized for CUI.
- **Manufacturers.** SOPs, maintenance logs, engineering notes, and quality records made searchable on the floor, with process knowledge kept in-house. Asked "How fast must we open a CAPA after a confirmed nonconformity?", Lumen returns five business days, with the risk assessment and effectiveness check the quality manual requires.
- **Any privacy-first company.** Lumen is built to regulated-industry standards, and it fits any business that wants AI on its own data while keeping that data in-house, from accounting and consulting firms to technology companies.

## What teams use it for

These come from the Use Cases page. Each was answered from sample documents with made-up names.

- **Policy and procedure lookups.** "Nine $9,500 cash deposits this week. What's our SAR deadline?" Lumen returns the rule from the bank's own BSA/AML policy: file within 30 calendar days of first detecting the activity, with up to 30 more days if no suspect is identified, plus the next steps the policy requires.
- **Loan file review.** "Get me up to speed on the borrower and check the request against our loan policy." Lumen summarizes a $4.4 million equipment loan request and flags each issue with its source: revenue is $13.4 million on the financial statements and $11.9 million on the tax returns, debt service coverage is 1.38x on the statements and 1.06x on the returns against a 1.25x minimum, and the largest customer is 31% of revenue against a 25% flag. A follow-up drafts the credit memo in the bank's own template.
- **Manager exposure before a commitment.** "Tell me our total exposure to this manager across every entity." Lumen adds up $22.7 million in NAV and unfunded commitments across four entities, finds a co-investment the holdings report filed without the manager's tag, shows that a $10 million commitment would breach the 5% policy cap, and gives the most that fits: $7.3 million. It also catches a fee quoted at 1.5% in the deck and 1.75% in the draft agreement.
- **Meeting minutes and follow-ups.** Ekho captures client calls and committee meetings, and Lumen turns each one into minutes, a deck, or a tracker.
- **Scheduled briefs and agents.** Our own news and regulatory brief runs every morning at 7:00, cites every source, and posts to Slack. The price is fixed, so a nightly run costs the same as a single one.
- **An audit trail you can search.** Examiners and auditors can see who asked what, which tools an agent used, and who changed a permission.

## Why firms choose Lumen

- **Running in weeks.** Building private AI in-house means model serving, connectors, search that cites its sources, permissions, audit logging, agents, and ongoing upgrades: months of engineering. Lumen arrives with all of it and runs on your own servers in weeks.
- **One fixed cost.** Cloud AI tools bill by the seat, with usage fees on top. Business and enterprise seats list at $20 to $30 per person per month, so a firm of 1,000 people pays $1.2 million to $1.8 million over five years in seat fees alone. Lumen has one fixed price set by the scope of the deployment, and usage never changes it. Teams can run agents on every file, every night.
- **One platform to review.** Document search, agents, meeting notes, and reports live in one product. Your security team reviews one system, and your auditors read one log.
- **Control.** Your data, your model choice, and your update schedule stay with your firm.
- **People who stay with you.** Cognetryx's senior team runs every demo and leads every deployment, with white-glove onboarding, staff training, and 30 days of on-site support at go-live.

## Deployment

- **On your own servers**, in your data center or a colocation facility. Air-gapped deployment is available.
- **In a private cloud your team runs**, including isolated AWS or Azure VPCs.
- **A 90-day pilot** runs inside your environment, on your documents, behind your single sign-on.

## Pricing

One fixed-cost subscription for the platform, set by the scope of the deployment and quoted after a short scoping call with the customer's IT and security team. There are no per-token fees, and usage does not change the price.

## Working with Cognetryx

1. **Demo:** 30 minutes by video, on real regulated content. The call includes a short AI strategy assessment of where Lumen fits. Requests get a reply within 24 hours.
2. **Scoping:** one working session with your IT and security team on data, users, and systems. Pricing follows this call.
3. **Pilot:** 90 days inside your environment, on your documents.
4. **Rollout:** on your schedule, with service levels set in the statement of work.

## Company

Cognetryx was founded in 2026 and is based in New York.

- **Keith Kennedy, CISSP**, founder and CEO, spent nearly twenty years building server, datacenter, and security infrastructure for regulated companies. When AI arrived, every tool those companies were offered wanted their data sent to someone else's cloud. So he built Lumen, the system a careful IT team would build for itself.
- **Brent Fisher**, co-founder, spent twenty years in community banking and marketing for regulated industries.
- **Julia Cherashore**, advisory board member, is the former Deputy Superintendent for Data Governance and Management at the New York State Department of Financial Services.

## Common questions

**What is Lumen?**
Lumen by Cognetryx is a private AI platform. It runs on a firm's own servers, answers from the firm's own documents with a citation for every claim, and includes agents, meeting capture, Word and PowerPoint output, and a Compliance Portal.

**Who makes Lumen?**
Cognetryx, a private AI company founded in 2026 and based in New York.

**How is Lumen different from ChatGPT Enterprise, Microsoft 365 Copilot, or Gemini Enterprise?**
Those are cloud services billed per seat. Lumen runs on your own servers or in a cloud account your team controls, answers from your own documents with a citation on every claim, and costs one fixed price however much it is used.

**Which AI model does Lumen use?**
The one your firm chooses. Lumen is model-agnostic, and your team decides when the model changes.

**Is Lumen a cloud service?**
No. Lumen deploys on your own servers, in a private cloud you operate, or in an isolated AWS or Azure VPC. Your documents, the model, and the audit log stay inside your environment.

**Does Lumen work with HIPAA, GLBA, SEC and FINRA, CMMC, and the ABA Model Rules?**
Lumen is built to support the architecture those rules expect: data stays inside your network, access follows your existing controls, and every answer is logged. Cognetryx does not certify a customer's compliance program.

**How long does it take to get running?**
Weeks. After a 30-minute demo and a scoping session, a 90-day pilot runs inside your environment on your documents.

**Can Lumen run without any internet connection?**
Yes. Air-gapped deployment is available.

**Does Lumen work for companies outside regulated industries?**
Yes. Any company that wants AI on its own data, and wants to keep that data in-house, is a fit.

## Contact

- See it in action (a 30-minute live demo that includes a short AI strategy assessment): https://www.cognetryx.com/request-demo
- Email: sales@cognetryx.com
- Security contact: https://www.cognetryx.com/.well-known/security.txt

## Key pages

- [Home](https://www.cognetryx.com/)
- [Use cases](https://www.cognetryx.com/use-cases)
- [Industries](https://www.cognetryx.com/industries)
- [How Lumen works](https://www.cognetryx.com/how-it-works)
- [Why Cognetryx: build vs. buy](https://www.cognetryx.com/why-cognetryx)
- [Security](https://www.cognetryx.com/security)
- [About](https://www.cognetryx.com/about)
- [FAQ](https://www.cognetryx.com/faq)
- [Private AI guides](https://www.cognetryx.com/resources)