- 1. Answer a question across one document or a hundred
- 2. Summarize a document and build a chronology
- 3. Compare documents and surface what is inconsistent
- 4. Make tracked edits directly in Word
- 5. Pull the key terms out of a document, or out of forty of them
- The thread that runs through all five
- Getting real value out of it
Category: Getting Started with NetDocuments
Most of the firms we talk to do not have the Legal AI Assistant switched on yet, and that is fine. It is a separate module you add to NetDocuments, part of the ndMAX suite, not something that comes turned on with the base platform. There is also a timely reason to look now, because the Legal AI Assistant has recently become fully available to firms in Canada. So before you decide whether it is worth adding, it helps to know what it actually does. And if you do already have it, treat this as a nudge, because plenty of firms who have paid for it are barely using it.
What sets it apart from a general AI chatbot is where it works. It runs on your own documents, inside your own tenant, under the same permissions and ethical walls you already have in place. You are not pasting confidential text into some outside tool. It reads the matter content you point it at, and it answers from that.
Here are five things it does well, and where each one actually saves time.
1. Answer a question across one document or a hundred #
This is the core of it. You select a document, a folder, or a whole set of matter files, and ask a plain language question. “What is the indemnity cap in this agreement?” “Which of these leases allow assignment without landlord consent?” You get an answer back in seconds, drawn from the documents you chose.
The part that matters for a lawyer is the citations. Every answer links back to the exact source passages it used, so you can click through and confirm it before you rely on it. You can also open the reasoning trace to see how it got there, and keep asking questions in the same thread as you narrow in. Treat the citation as the point, not a footnote. The answer is a shortcut to the source, not a replacement for reading it.
2. Summarize a document and build a chronology #
Hand it a long examination for discovery transcript, a stack of correspondence, or a dense agreement, and ask for a summary. More usefully for litigation, ask it to build a chronology. It will pull the dated events out of the material and lay them out in order, which is one of those tasks that eats an afternoon when you do it by hand.
This works best as a first draft. It gets you a structured starting point in minutes, and you refine from there.
3. Compare documents and surface what is inconsistent #
Point it at two versions of a contract, or a set of agreements you want checked against a standard, and ask what differs. It is good at finding the needle in the haystack: the one clause that changed, the defined term used two different ways, the provision missing across a batch. For anyone doing volume review or due diligence, this turns a slow manual pass into a targeted one.
4. Make tracked edits directly in Word #
This is the newer, agentic side of the assistant, and it is the one that surprises people. Working inside Microsoft Word, you give it a plain language instruction. “Change the notice period to 30 days throughout.” “Make the defined terms consistent.” “Tighten this recital.” It makes the edits as tracked changes, right there in the document, so you review and accept or reject them exactly like a colleague’s redline. Nothing leaves your normal workflow, and you stay in control of every change.
5. Pull the key terms out of a document, or out of forty of them #
Ask it to extract the specific data points you care about and lay them out in a table. Parties, effective dates, renewal and termination dates, notice periods, governing law, dollar figures. Do it for a single agreement, or run it across a folder of them to abstract a whole portfolio of leases or contracts at once. What used to be a student spending a week on a spreadsheet becomes a first pass you review and correct. As always, the citations let you check each figure against the source before it goes anywhere.
The thread that runs through all five #
The reason these are worth trusting is the same reason to bother with any of it. The AI is brought to your content, rather than your content being handed to some outside tool. Everything stays inside your NetDocuments tenant. Answers respect the security you already have, so someone walled off from a matter cannot pull it up through the assistant. Your documents are not used to train public models, and administrators can layer on data loss prevention rules to control what sensitive content the AI is allowed to touch.
Two honest caveats. It works on the documents you give it, so it is not a case law research tool. And like any AI, it can be confidently wrong, which is exactly why the citations are there. Check them.
Getting real value out of it #
Whether you are deciding whether to add it or you already have it and it is gathering dust, the same thing separates the firms that get value from the ones that do not. It is not the technology. It is matching the assistant to the work your team actually does day to day, and setting it up so people trust the output enough to build it into their routine. That is the part worth being deliberate about.
If you are running NetDocuments and wondering whether the Legal AI Assistant is worth adding, or you already have it and want your team actually using it, that is what we do. Reach out at hfsm.ca/contact.
