LexisNexis Protégé vs Harvey vs CoCounsel: What's Missing From All Three

The legal AI market just passed a threshold. Within two weeks in February 2026, LexisNexis replaced its entire Lexis+ AI platform with a new product called Protégé. Thomson Reuters acquired Noetica to bolster CoCounsel. Harvey AI quietly crossed 100,000 lawyer-users and is on track for roughly $190 million in annual recurring revenue. And according to the latest surveys, 98% of AmLaw 200 firms now use AI tools in some capacity.

The arms race is real. The budgets are real. And every one of these platforms has the same blind spot.

They can't see your firm's own documents.

What Each Tool Does Well

Let's give credit where it's due.

LexisNexis Protégé launched on February 24 with 300+ pre-built legal workflows — contract review, deposition prep, regulatory analysis, citation checking — all running on top of LexisNexis's massive case law and statutory database. It's the most polished attempt yet to turn legal research into something that feels like a conversation. If your work revolves around public legal sources, Protégé is a serious upgrade. Pricing sits in the $500–$1,000+/user/month range depending on the bundle.

Harvey AI has taken a different path. Rather than building on a legal publisher's database, Harvey has embedded itself directly into Anthropic's Claude models, training on legal reasoning patterns and building deep integrations with firm workflows. At roughly $1,000–$1,200/lawyer/month with a 20-seat minimum, Harvey is firmly positioned as an enterprise tool for the AmLaw 100. It's fast, handles multi-step legal reasoning well, and firms like A&O Shearman and PwC Legal have committed publicly.

Thomson Reuters CoCounsel — now reinforced by the Noetica acquisition on February 10 — occupies the more accessible end of the market at $250–$500/user/month. It's tightly integrated with Westlaw and Practical Law, which gives it a built-in user base of hundreds of thousands of lawyers who already live inside the Thomson ecosystem. CoCounsel's bet is that the best AI is the one that lives inside the tools you already use.

Three different strategies. Three well-funded companies. One shared assumption: that the documents that matter most are the ones they already have.

The Blind Spot They All Share

Here's what none of these tools touch: the 20 years of work product sitting on your firm's own servers.

Every mid-to-large law firm has a document heritage — hundreds of thousands, sometimes millions, of files accumulated over decades. Briefs. Memos. Engagement letters. Deal closing binders. Email chains with nuanced strategy discussions. Due diligence reports. Precedent clauses refined over hundreds of transactions. All of it sitting in a DMS, an email archive, a network drive, a legacy server that nobody has migrated yet.

This is the institutional knowledge that makes a 200-lawyer firm different from a solo practitioner with a Westlaw subscription. It's the reason clients pay $800/hour — not because the lawyer can find a statute faster, but because the firm has done this exact deal before, in this exact jurisdiction, with this exact counterparty.

Protégé doesn't index it. Harvey doesn't index it. CoCounsel doesn't index it.

These tools search LexisNexis databases, or Westlaw databases, or the open web. They can summarize a contract you upload. But they cannot answer the question that matters most to a practicing lawyer: How did we handle this last time?

That's not a feature gap. It's an architectural limitation. SaaS platforms can't deeply index a firm's internal document universe without the firm sending that data to an external cloud, which raises exactly the kind of privilege and confidentiality concerns that — as the Heppner v. United States ruling recently demonstrated — courts are now scrutinizing in real time.

Why This Matters More Than You Think

The immediate cost is invisible: it's the senior associate spending three hours searching the DMS manually for a precedent brief she knows exists, or the partner re-drafting a motion from scratch because the last version is buried in a departed colleague's email archive.

But the strategic cost is larger. Firms are paying $500–$1,200 per user per month for tools that access the same public databases available to every other firm using the same tool. The competitive advantage is zero-sum at best. When everyone has Protégé, nobody has an edge from Protégé.

The edge comes from what your competitors don't have: your firm's accumulated knowledge. The precedent bank built over decades. The clause library refined through a thousand negotiations. The internal research memos that contain the real reasoning — not the sanitized version in the published opinion.

That knowledge is currently dark. It's not searchable by any of these tools. And the longer firms invest exclusively in SaaS legal AI without addressing their own document heritage, the wider the gap grows between what AI could do for them and what it actually does.

What Proprietary AI Gets Right

The alternative isn't complicated in concept, though it requires a different architecture. Instead of sending your documents to a vendor's cloud, you deploy AI on your own infrastructure — your private cloud, your on-prem servers, your existing security perimeter.

The AI indexes everything: the DMS, the email archive, the legacy servers, the network drives. It builds a searchable, queryable knowledge base from the firm's entire document heritage. And because it runs inside your walls, there's no data leaving the building. No third-party retention. No terms of service that could compromise privilege.

This solves the three problems the major platforms leave open:

1. Internal document access. The AI can actually answer "How did we handle this last time?" because it has indexed the answer.
2. Data sovereignty. The firm owns the model, the index, and the infrastructure. No vendor has access to the data. The deployment would pass the test Judge Rakoff applied in Heppner.
3. Competitive differentiation. Two firms can buy the same Harvey license. They cannot buy each other's 25 years of deal memos.

The Question Worth Asking

The next time a vendor pitches you a legal AI platform, ask one question: Can it search our internal documents without our data leaving our infrastructure?

If the answer is no — and for Protégé, Harvey, and CoCounsel, it is — then you're buying a better version of Google Scholar, not a system that understands your firm.

The firms that figure this out first will have a structural advantage that SaaS tools can't replicate. The firms that don't will keep paying more for tools that make every firm look the same.

At RAGbase Legal, we build proprietary AI systems that index a firm's full internal document heritage and deploy on the firm's own infrastructure — zero data retention, full code ownership. If you want to see what it looks like on your own documents, we run a free 3–5 day proof of concept. No pitch deck. Just your data, your servers, working results. Reach out at hadi@ragbase.ai.