Private Intelligence by VerdictIQ
An AI assistant that answers questions about your firm’s own documents, running entirely on a computer in your office. No cloud. No third parties. Nothing your clients told you in confidence ever crosses the door.
Cloud AI is off limits for the work that matters most. Private Intelligence lets you answer that question with one word: nowhere. The model lives on your machine, and the answer is produced on your machine.
Why firms trust it with privileged material
Data Residency
Every answer is produced on your own machine, by a model that lives on your machine. The system refuses to send your document text to any outside service.
For you: answer "where does our data go" with one word. Nowhere.
Ethics Wall
Tag a document to a client matter and scope a question to it. The assistant only uses that client’s material.
For you: an ethics wall for matters tagged on upload, so one client’s file never surfaces in another’s answer.
Applies to documents tagged on upload. Bulk indexed folders are firm wide.
Grounded Output
Every answer is traced to a source in your documents. If the answer is not in your files, it says so instead of inventing one, and it is built not to give legal advice.
For you: grounded, cited answers or an honest "I could not find that."
It reduces hallucination. It does not turn a local model into a senior associate.
Input Security
Documents are treated as untrusted data, so a file with hidden instructions cannot hijack the assistant or pull it off its rules.
For you: an opposing counsel PDF cannot quietly turn your assistant against you.
Mitigated, not eliminated. Curating what enters the index is the real control.
App Security
The app rejects requests that do not come from its own screen, closing the browser based tricks that target software on a local machine.
For you: a stray website a staffer visits cannot reach in and touch your documents.
Transparency
The product documents plainly what is protected, that setup needs internet but daily use does not, and where the limits are. No security theater.
For you: a vendor you can put in front of a risk committee without surprises.
How It Works
STEP 01
Point it at your files. They are read, chunked, and embedded locally into a private index on your own hardware. Nothing is uploaded anywhere.
STEP 02
Ask in plain language and scope the question to a client matter or a department. The question and the documents never leave the machine.
STEP 03
A local model answers using only your retrieved documents, with citations, or tells you honestly when the answer is not in your files.
See It
A cited answer drawn only from the firm’s own indexed files, with the relationships it used. This is the actual VerdictIQ interface, shown here with synthetic demo documents.
Who It’s For
Independently Reviewed
Every security control was hardened through a structured audit, then cross checked fix by fix by three independent AI reviewers until each one was clean. The protections are locked in by an automated test suite of more than forty tests that runs with the network fully disabled, proving the offline promise rather than just claiming it.
The Deliverable
This is not a login to one more cloud tool. We spec the hardware, provision the model and software, and deliver a working appliance that runs behind your door. Setup is a one time step, and from then on your firm asks questions of its own files with nothing leaving the building. Every engagement starts with a consultation, not a checkout page.
FAQ
Nowhere. Documents are indexed and answered on your own hardware. The system refuses to send document text to any outside service, so the answer to "where does our data go" is simply: it does not.
At query time, yes. A one time setup needs internet to install the software and the model. After that, daily use runs with no outbound calls, and you can confirm it with the network disabled. The offline guarantee is about operation, not the initial provisioning.
The local index holds your document text and embeddings in plaintext on the machine, so at rest protection comes from the appliance’s full disk encryption and keeping it locked when unattended. Nothing is stored off site, and we are explicit about this so your risk committee can plan for it.
No. It answers questions about your own documents and is built not to give legal advice. Using it creates no attorney client relationship.
For documents tagged to a matter on upload, no. The assistant only uses that matter’s material. Documents added through a bulk folder index are treated as firm wide, so matter tagging on upload is how you keep the wall between clients.
ChatGPT sends your text to a third party cloud. Private Intelligence runs on your hardware and never does. GateKeeperAI captures and qualifies new leads at the front door, while Private Intelligence works behind the door on the privileged files your firm already owns.
We spec and provision it. A GPU workstation for a small firm, a server for a larger one, delivered ready to run behind your door. The hardware is part of the deliverable.
The Full AI Stack
Captures and qualifies every new lead across calls, chat, and forms, then books the consultation.
Makes your firm the one ChatGPT, Perplexity, and Google AI Overviews cite when prospects ask AI for a lawyer.
Secures the firm’s own knowledge: private AI over your privileged files, on your hardware.
On your hardware. Behind your door. Answers from the files your firm owns. Book a call and we will scope what a private AI appliance looks like for your firm.
Private Intelligence by VerdictIQ does not provide legal advice and creates no attorney client relationship. The offline guarantee applies to query time operation; a one time provisioning step installs the software and models.
