NetDocuments

What it is

NetDocuments is the cloud-native half of the legal document management duopoly — the other half being iManage. It is where more than 7,000 law firms, corporate legal departments, and public sector organizations across 90+ countries file work product, with connections to 150+ other technologies. Unlike iManage, it has no on-premises heritage to carry: it was built as a hosted system and has never sold a server-room edition of itself.

In 2026 it also became a consolidator. On January 12, 2026 it closed the acquisition of eDOCS from OpenText for $163 million in cash, taking the product portfolio, the customer contracts, and the eDOCS team; OpenText had disclosed the unit at roughly $30 million of annual revenue in its fiscal year ended June 2025. That follows the 2022 purchase of Worldox. The two acquisitions are being handled differently, and the difference is the part buyers should read: Worldox on-premises reaches end-of-life on December 31, 2026, while NetDocuments has said eDOCS on-premises has no end of life planned.

Why it shows up in Legal Ops stacks

• ndConnect Enterprise is an MCP server, and it is a tier — not a feature. The Enterprise tier lets AI agents autonomously search and retrieve documents on behalf of an authenticated user; Claude, ChatGPT, Perplexity, Copilot Studio, and Gemini Enterprise connect through it. Standard ndConnect, available to all customers, is the manual version: a human selects documents and loads them into an approved partner app. Both are called ndConnect. Only one of them is an agent integration.
• Governance is enforced at the connection layer, not by the AI vendor. Existing permissions, ethical walls, and DLP rules apply on every call, access is scoped to the authenticated user, and documents stay in NetDocuments rather than being copied into a vendor index. Administrators opt in to each AI tool individually through Security Center, so “we enabled MCP” is a per-tool decision rather than a switch.
• Smart Answers put natural-language search over the firm’s own repository. Announced March 4, 2026 and rolling out from March 31 to ndMAX Enterprise customers, it answers questions with citations back to filed documents, inside NetDocuments rather than through a separate contract. Direct connectivity to ChatGPT and Claude expanded on April 1, 2026, also for ndMAX Enterprise.
• ndMAX is three things, and they get bought as one. A legal AI assistant that queries one file or hundreds; an AI app builder for low-code applications scoped to a practice area or jurisdiction; and ndMAX Studio, a catalog of ready-made apps covering translation, summarization, and playbook analysis.
• The model layer is Microsoft’s. ndMAX runs on Microsoft Cognitive Services and Azure OpenAI models. Inputs and outputs are not retained past the exchange and are not used for training, and NetDocuments holds an agreement with Microsoft that keeps customer content out of abuse-monitoring systems.
• Legal AI vendors index it rather than replace it. Harvey, Legora, Definely, DraftWise, Luminance, and vLex Vincent AI all ship NetDocuments connectors, which makes “what can our AI see?” a question about NetDocuments configuration.

Pricing

NetDocuments does not publish a price for ndMAX. Its own ndMAX product page returns Custom Pricing in the page’s structured data, and every route ends at a demo request. Any per-user AI figure in circulation — including the widely repeated $15–25 per user per month for the add-on — comes from resellers, consultants, and comparison sites, not from the vendor. Use it as a negotiating anchor; do not budget from it as a rate card.

• Reported DMS band: roughly $40–65 per user per month for a legal-grade configuration, per reseller and consultant reporting current as of August 2026. Entry editions are quoted nearer $20, and UK channel partners land around £35–55 per user per month.
• Implementation is a separate invoice — commonly $1–5K for small and mid-sized firms and $5–20K+ for enterprise deployments, depending on migration scope.
• The tier is the real cost driver. Smart Answers, direct ChatGPT and Claude connectivity, and ndConnect Enterprise all name ndMAX Enterprise as the prerequisite. The gap between an ndMAX quote and an ndMAX Enterprise quote is the gap between a chat assistant and an agent-addressable repository.

Best for

Law firms and in-house legal teams that want matter-centric filing with no on-premises estate to migrate, and legal ops leads who need matter-level access control to survive contact with an AI assistant.

The case NetDocuments wins outright: you want a governed connection between twenty years of filed work product and Claude, ChatGPT, or Copilot without exporting that corpus to a vendor’s index, and you want to admit one AI tool at a time by administrator opt-in rather than issuing a blanket API credential.

Alternatives

• iManage — the other half of the duopoly, reported at roughly $50–75 per user per month on an approximately 10-seat minimum with implementation at $10–25K+. Pick it when you have an on-premises or hybrid estate to accommodate, or when matter-level ethical walls in Security Policy Manager are a written procurement requirement. Its MCP Server reached general availability on May 14, 2026 and is not gated behind a premium AI tier — compare the tier you would actually buy, not the checkbox.
• Microsoft SharePoint / Microsoft 365 — pick it when you already pay for E3 or E5, matter volume is low, and nobody is asking you to prove need-to-know access at the matter level. What you give up is the legal metadata model and ethical-wall enforcement, and you will rebuild both badly.
• DeepJudge — the fastest-growing entrant in the segment, with revenue growth above 500% year over year and a $41.2M Series A led by Felicis with Coatue at roughly $300M, naming Freshfields, Holland & Knight, and Cozen O’Connor. It does not replace the DMS. Pick it alongside NetDocuments when the problem is search across the DMS, email, intranet, and client portals rather than filing.

Watch-outs

• “ndConnect” names two products with different capabilities. A quote that says ndConnect without saying Enterprise buys the manual tier, where the integration is a person selecting files. Guard: make the order form name ndMAX Enterprise, and list Smart Answers and ndConnect Enterprise as separate line items with the entitlement written out.
• There is no published AI price to hold the vendor to. Guard: require the AI add-on quoted as its own per-user line separate from the DMS line, and benchmark the total against an iManage quote at the same seat count. In a two-vendor market, cross-quoting is the only pricing pressure available.
• Two acquired products, two different commitments. Worldox on-premises ends December 31, 2026; eDOCS has no announced end of life. Guard: if you run either, get the support horizon and migration-assistance terms into the renewal document — a “no end of life planned” line in a press release is not a contractual term, and the Worldox timeline shows what the other outcome looks like.
• MCP grants exactly what the authenticated user already has. Years of loose need-to-know configuration surface as an AI returning privileged material to the wrong lawyer; years of over-tight configuration surface as an AI that finds nothing. Guard: audit security classes and workspace access on the matters in scope before the first pilot, and use the Security Center opt-in to admit one AI tool at a time.
• The inference layer belongs to Microsoft, which your policy may care about. Guard: if your data-residency or model-provenance policy names approved models or regions, confirm which Azure region and model version your tenant runs before signature. “Your data is not used for training” answers a different question than “where does inference happen.”