It's been quite a few weeks in legal research technology. If you blinked, you might have missed Anthropic launching Claude for Legal with a dozen practice-area plugins and over 20 MCP Connectors. Or Legora acquiring legal research startup Qura and Graceview, a regulatory horizon scanning platform. Or Clio pushing deeper into research with its vLex integration through Vincent AI and the new Studio workflow builder.
The headlines all land in roughly the same place: AI is going to transform legal research. Automate it. Democratize it. Maybe even replace the people who do it.
I've spent my career at the intersection of legal research, knowledge management, and innovation as a librarian, a lawyer, a Knowledge and an R&D Director, and now leading Knowledge and Research Services at Harbor. So I've seen a fair number of "this changes everything" moments in legal technology.
And I want to be clear: some of what's happening right now genuinely does change things. But not in the way most of the commentary suggests.
Let's Start with What's Actually Exciting
I'm not going to be the person standing at the back of the room with her arms crossed pretending that this is smoke and mirrors. These are real products doing real things.
Claude for Legal's MCP Connectors pull together impressive data sources into a single conversational interface. If you've ever had six browser tabs open across three different platforms trying to triangulate an answer to a research question, you'll immediately understand why that matters.
Legora's aOS is promising end-to-end agentic work across research, drafting, and review. Vincent AI from vLex delivers cross-jurisdictional research across over 100 countries. These are serious teams building serious products, and some of them, particularly the access-to-justice integrations with CourtListener, Courtroom5, and Descrybe.AI, are addressing problems our industry has been ignoring for decades.
But there's a distinction here that is easy to miss in the excitement, and a brilliant piece from Legaltech Hub's Stephanie Wilkins nailed it: retrieval of some relevant authority is fundamentally different from comprehensive search of all relevant authority. A Deep Research output from Claude's Thomson Reuters Connector gives you a synthesized memo drawing on relevant sources.
What it doesn't give you is a search result list you can inspect to make sure nothing was missed. It doesn't give you direct access to Westlaw or Practical Law. It doesn't give you a standalone citator you can independently verify.
That's not a criticism. It's a description of what the tool does. And knowing the difference matters enormously.
Here's What I Think the Market is Getting Wrong
The instinct from much of the commentary is that more AI research tools means less need for research professionals. I think the opposite is true.
When a firm had one research platform, the research question was bounded. You learned the system, you understood its editorial architecture, you knew what it covered and where its gaps were. The expertise was deep but the environment was relatively contained.
Now consider what a firm's research landscape might actually look like soon: Harvey for workflow automation, Legora for agentic research and drafting, Claude with MCP Connectors pulling from CourtListener, Midpage, and Descrybe, Vincent AI for cross-border work, and still a Westlaw or Lexis subscription for comprehensive coverage and citation validation.
That is not simplification. That is an ecosystem. And if I've learned anything from spending over a decade inside global law firms, it's that ecosystems don't manage themselves. They need people who understand the whole picture - what each tool does well, where each one has gaps, how to combine them intelligently, and when the output needs to be independently verified against authoritative primary sources.
The researcher's role isn't being automated away. It's being elevated. The question is no longer just "find me the relevant case law." It's "orchestrate these six tools, evaluate the completeness and reliability of what each one returns, identify the gaps, fill them, validate the currency and treatment of every authority, and give me a confident answer."
That's a harder job than what came before. Not an easier one.
What I'm Seeing From Inside Firms
I started my career as a librarian. I've been a practising lawyer, a Knowledge Director, a Head of Innovation and an R&D Director at a Magic Circle firm. Now I'm working across dozens of firms through Harbor. And what I consistently see is that the most effective teams aren't the ones trying to figure out whether AI replaces their researchers and librarians. They're the ones investing in those people and giving them these new tools to work with.
A skilled legal information professional brings something that no current AI tool can replicate: the judgment to evaluate whether a research result set is actually complete. The instinct to spot when an AI synthesis has missed something. The understanding of how different sources are editorially structured and where their jurisdictional boundaries lie. And the knowledge of which tool is the right one for a given question because right now, no single tool is the right one for every question.
These are the people who are also best positioned to evaluate new tools when they land, train lawyers on how to use them properly, and build the governance frameworks that ensure AI-assisted research meets the firm's quality standards. If innovation can't be a side gig (and I've been saying that for years) then neither can the work of making sure these tools are deployed thoughtfully.
Where Does This Leave Us?
The legal research landscape is changing fast. But faster is not the same as complete. Accessible is not the same as reliable. And powerful is not the same as correct.
The human layer, the researcher who knows when to trust the synthesis and when to verify it, who understands the architecture of the sources, who can evaluate completeness rather than just relevance, is not a legacy cost to be managed down. It's the thing that makes all of these remarkable new tools actually work for the practice of law.
At Harbor, we're helping our clients lean into this moment rather than react to it. We're working with firms to integrate these tools thoughtfully, build research workflows that pair AI capability with professional judgment, and make sure the promise of these technologies is realised without the risks that come from treating them as more than they currently are.
The tools have never been better. The need for the people who know how to use them has never been greater.
- Research
- AI
- Knowledge management
- Show all 7
