The first move in the legal DMS market’s AI repositioning
Neil Cameron, lead analyst
NetDocuments today (14 May) will announce what CEO Josh Baxter and chief product officer Dan Hauck described as a reimagined platform built around a new substrate they are calling the Legal Context Graph. Private preview opens today for customers on the Enterprise AI tier, with public preview to follow. It is a substantial announcement, and one that deserves to be taken seriously – both for the ambition behind it and for what it signals about where the legal AI conversation is heading.
The argument, stripped of slideware, is that AI in legal work is only as useful as the context it can reach, and that context in a law firm is scattered across matters, documents, communications, timelines, and institutional knowledge held in people’s heads, all bound by permissions and ethical walls that cannot slip. NetDocuments’ response is to construct a typed, traversable graph spanning three tiers – global (industry and expertise), matter (case-specific), and document – and to feed it into an AI context engine that any model, first-party or third-party, can draw on. The live demo, walked through by director of product Nate Ruiz, showed the graph manifesting as cross-matter natural-language search, an automatically assembled matter overview with extracted parties and dates, version-difference summaries, and an in-Word drafting panel. The latter pulled a freshly filed expert report into a Markman reply brief without anyone having to tell it where to look. It was a polished and well-staged demonstration.
Where the substance sits
The architectural direction is one that several vendors are now travelling along, and the broader convergence is significant in its own right. Caroline Hill’s 14 April exclusive on these pages flagged iManage CEO Neil Araujo’s description of his own forthcoming announcement at ConnectLive 2026 – to be unveiled in Chicago in May and London in June – as a layer that sits on top of the data to provide context for AI, and as significant as the launch of iManage Cloud. Whatever the iManage announcement is ultimately called, the shape of it is unmistakable: two of the legal DMS market’s leading vendors arriving, within weeks of each other, at the same conclusion about where the structural value lies, and reaching for the same headline noun – context – to describe it. The DMS sits in a privileged position to be the trust-and-governance substrate for legal AI, because it holds the documents, the matter metadata, the permissions, and the editing history. That iManage and NetDocuments have both arrived at versions of this conclusion is the more important story; the specific differences between their implementations will become clearer as both are deployed in customer environments through the rest of the year.
What is visibly impressive about the NetDocuments announcement is the breadth of first-party AI surface area on display. The in-Word side panel grounded in matter context, the automatic version-difference summaries, the matter overview with a ‘completeness’ indicator, and the cross-tenant Smart Search are concrete applications, not abstractions. The Word co-authoring integration – the document never leaves NetDocuments – is a genuinely good piece of engineering, and the partnership with Microsoft behind it is substantive. The polish reflects the 39 design studies involving 1,500 participants that Hauck cited as informing the new interface.
Hauck confirmed that the graph schema draws on SALI’s Legal Matter Specification Standard, of which NetDocuments is an Alliance member, alongside platform-specific adaptations. That is the right architectural choice, and worth highlighting. A context graph keyed to an open industry taxonomy is, in principle, interoperable with any client or vendor system that also tags to LMSS. The novelty here is not in the taxonomy – SALI did that work – but in the graph-construction and retrieval layer built over it, with permissions and ethical walls intended to be enforced at retrieval time rather than only at document-open time.
Questions for the road ahead
The “first-ever legal context graph” framing is the kind of claim that vendor launches tend to make, and Hauck’s own acknowledgement that the company has been doing document enrichment for some time – and has now “dramatically expanded” it and added graph capabilities at the matter and global levels – places this announcement on a continuum of accumulated investment. The platform is model-agnostic in the conventional sense (Anthropic this week, OpenAI next month, several models depending on task), which is increasingly table stakes. The eDOCS acquisition, completed in January and following the Worldox playbook of three years earlier, Baxter candidly clarified, is about distribution rather than the technology underpinning this announcement.
The most telling question I put to the team – and one that will matter increasingly as firms scale GenAI-assisted drafting – is how a supervising partner reviewing an associate’s AI-assisted work knows what context the Legal Context Graph fed into the model, and whether that is auditable on a per-prompt basis. Hauck pointed to the thinking traces that modern reasoning models expose, and to the platform’s existing governance architecture for retention, but this area warrants closer scrutiny. This is a live issue for the whole industry rather than a NetDocuments-specific issue, and one I shall be watching closely as the private preview progresses: whether the platform can produce, on a per-prompt basis, a structured record of which documents, clauses, and matter context entered the model’s context window, under whose permissions snapshot, against which model version, and retained for how long. If that record is there, NetDocuments will have moved meaningfully ahead of where most of the vendor market currently sits on the validation question. I look forward to seeing it demonstrated in subsequent briefings.
What this means for firms
For firms already on NetDocuments, the upgrade path is unusually low-friction by the standards of platform reimaginings at this scale: Hauck confirmed no migration is involved, with users able to toggle between the existing platform and the new experience at their own pace. That alone will accelerate adoption among customers who would otherwise have stalled at the change-management hurdle. For the wider market, an important question is what happens to third-party legal AI vendors as both DMS incumbents position themselves as the natural context source for AI inside law firms. Vendors who have built horizontal AI products on top of customer document estates now find the substrate beneath them developing into something more active. That is a structural shift worth tracking through 2026.
This is a serious piece of work. As more detail emerges over the coming months, we shall learn a great deal about how the two leading legal DMS vendors intend to compete on the same architectural ground – and in that comparison, the shape of the 2026 legal DMS market will begin to take form.
