Our Take
The real story is not that AI landed in Word this month, but that legal tech is fracturing into two tiers: vendors building atop foundation models (Anthropic, Microsoft) and specialists solving problems that require jurisdiction-specific data and institutional memory (Eudia, Stretto, ViClarity).
Why it matters
In-app AI is commoditizing contract review, which means legal ops teams now face two parallel infrastructure decisions: whether to adopt these embedded tools at all, and if not, whether to build custom stacks atop the same foundation models. The winners will be those solving unsolved problems that generic AI cannot reach alone.
Do this week
Legal operations leaders: map your contract review workflow against Claude for Word and upgraded Copilot capabilities by end of week so you can identify which use cases genuinely require your in-house precedent and which ones the foundation models can handle unsupervised.
Anthropic and Microsoft both shipped legal-focused AI in April
Claude for Word and Microsoft's upgraded Copilot landed in the same week, each with tracked-changes output and explicit targeting of contract review and due diligence workflows. Both vendors positioned their tools as drop-in augmentation for lawyers already working in Word, requiring no new application layer or data migration.
The speed of the market registered immediately. By the time the April product table was compiled, Anthropic had already released Claude for Legal, which will appear in the May update.
Specialized vendors are racing to own the parts AI cannot yet reach
While foundation models are colonizing contract review, three other legal tech launches in April targeted problems that demand domain specificity and structured data. Eudia's Expert Digital Twins addresses institutional knowledge capture in legal operations, a problem that has resisted tooling for decades because it lives in individual experience rather than datasets. Stretto's Research Suite goes deep into bankruptcy-specific precedent, where generic legal training data proves insufficient. ViClarity's Reg Monitor connects regulatory horizon scanning to auditable compliance workflows, requiring jurisdiction-specific infrastructure and GRC integration.
The separation is stark: foundation models solve broad problems at low friction. Everything else demands specialization, data control, and audit trails. Legal teams will now operate across both categories simultaneously, which means different vendor selection criteria for each.
Audit your contract review workflows first
Start with the work that foundation models are already eating. Contract review and due diligence are the clearest candidates for Claude or Copilot because they rely on pattern matching and precedent reasoning that scaled language models handle well. If you are currently paying vendors or burning internal capacity on these tasks, test the embedded tools immediately. They will likely be cheaper and faster.
Once that ground is clear, map the rest of your legal operations stack: institutional knowledge management, regulatory tracking, compliance workflows, and domain-specific research. These are the problems where specialized vendors now have room to operate. You will need both categories, but the question is no longer whether to use AI in Word. It is which problems you stop outsourcing because AI can do them, and which ones you invest in deeper tools to solve.