Legal AI moat equals switching costs

Legal AI looks less like a software moat race and more like a distribution and workflow land grab. The products are converging quickly, buyers can now get five seat pilots and even six month terms, and large firms often choose Harvey because clients ask for it by name, not because it is clearly superior. What sticks is training, approvals, integrations, and the hassle of ripping out a tool once a practice group has built habits around it.

• At large firms, both Harvey and Legora usually land as small practice group deployments rather than true firm wide platforms, because enterprise scale licensing is too expensive relative to the benefit. That makes the real lock in local champion behavior, user training, and license management, not deep technical differentiation.
  1 sacra
• The side by side product gap is real but narrow. Harvey wins more on brand, client pull, and US legal mindshare. Legora is stronger on workflow structure, collaboration, international use, and knowledge vaults. Buyers increasingly keep relationships with both, which is a sign of weak exclusivity.
  1 sacra 2 sacra 3 sacra
• The deeper reason the moat looks thin is that frontier models have already commoditized a lot of legal reasoning. That is pushing the market toward either workflow specific tools like contract review and drafting, or incumbents like Thomson Reuters, Clio, Ironclad, and Icertis that already own the system where legal work happens.
  3 sacra 5 sacra 6 sacra

The category is heading toward fewer general purpose winners and more value accruing to whoever owns daily workflow, internal knowledge, and the surrounding system of record. If Harvey or Legora become hard to remove, it will be because they are wired into document stores, approvals, and team habits, not because the underlying model capability stays unique for long.