Artificial intelligence is everywhere in legal technology right now, and law firms are right to ask hard questions before adopting it. Who can see our data? What is the AI actually searching? Could it pull in inaccurate information from the internet? With NetDocuments AI — whether you’re using the AI Assistant or Pattern Builder Max — the answers are straightforward: your data stays yours, and the AI only works with what’s in your firm’s repository.
Your Documents. Only Your Documents.
When an attorney runs a query through the NetDocuments AI Assistant or uses Pattern Builder Max to extract information and classify documents at scale, the AI is working exclusively against the documents stored in that firm’s NetDocuments cabinet. It is not pulling results from the open web, it is not referencing other firms’ repositories, and it is not sending your documents to a shared pool where other users could benefit from or access your content.
This is a meaningful architectural distinction. Many general-purpose AI tools — the kind attorneys might be tempted to use on their own — are designed to search broadly across the internet or draw on publicly available information to generate answers. That approach introduces real risk for law firms: confidentiality concerns, inaccurate results sourced from unreliable web content, and the possibility that client information gets submitted to an external service that logs or retains it.
NetDocuments AI does not work that way.
No Hallucinations From the Open Web
One of the most talked-about risks with AI tools is hallucination — when the AI confidently states something that isn’t true. In many cases, this happens because the AI is drawing on broad training data or real-time web search results of uncertain quality. If an attorney asks a general AI tool about a case, a statute, or a legal standard, there is a real possibility the tool will produce a plausible-sounding but fabricated or outdated answer.
NetDocuments AI avoids this problem by design. Because it queries only the documents in your firm’s repository, it can only surface what is actually there. If the answer isn’t in your documents, the AI won’t make something up from a web search. That constraint — which might sound like a limitation — is actually one of the most important features for a law firm context, where accuracy and source integrity are non-negotiable.
Your Data Doesn’t Train the Model
A concern we hear from firms evaluating AI tools is whether their documents and queries are being used to train or improve the underlying AI model. This is a legitimate question, and the answer with NetDocuments matters: your firm’s documents and prompts are not used to train the AI. When an attorney uses the AI Assistant to ask a question or Pattern Builder Max to process a document set, that activity stays within the firm’s environment. It is not feeding a shared model that other organizations could indirectly benefit from.
This is different from the way many consumer-facing AI tools operate, where user interactions may be used to improve model performance over time. For law firms handling sensitive client matters, that distinction is important — and it should be part of any due diligence conversation when evaluating AI tools for practice use.
What This Looks Like in Practice
With the AI Assistant, attorneys can ask natural-language questions about their documents — finding agreements that contain specific provisions, summarizing a matter’s files, or locating relevant precedents from past work — and every result is grounded in documents the firm already controls and has already vetted.
With Pattern Builder Max, the AI can analyze documents at scale to extract metadata, classify document types, identify key terms or dates, and perform tasks that would otherwise require hours of manual review. All of this happens within the firm’s own NetDocuments environment. The processing is not being handed off to an external AI service that operates outside the firm’s data governance structure.
Access Controls Still Apply
NetDocuments AI also respects the security structure the firm has already established. If an attorney only has access to certain cabinets or workspaces, the AI Assistant honors those same permissions. The AI does not have broader access than the user running the query. Ethical walls, matter-level security, and cabinet restrictions all remain in effect — the AI works within the security framework, not around it.
The Bottom Line
NetDocuments AI gives law firms the efficiency benefits of artificial intelligence without the data privacy risks that come with general-purpose tools. Your documents stay in your repository. Queries run against your content only. Your data isn’t used to train a shared model. And the results reflect what’s actually in your files — not what an AI guessed from the open internet.
If you’re evaluating AI tools for your firm and have questions about how NetDocuments AI fits into your practice, Optiable has been implementing and supporting NetDocuments at law firms since 2010. Book a consultation at go.oncehub.com/optconsult.