Legaltech Hub

When a new AI model appears, the temptation is to use the flagship version for everything. It feels like the safest option. In reality, it often means paying premium prices for routine work, while overlooking where additional reasoning actually creates value. This article explains how the Claude model family fits different types of legal work, how reasoning effort influences both quality and cost, and why risk provides the clearest framework for making those decisions.

Perhaps the greatest lesson we've learned is that AI strategies should be designed around people, not technology. Too many organizations begin by asking which AI platform they should deploy. A better question is how their people naturally want to work. When organizations start with their talent rather than their technology stack, better decisions follow. Employees are happier, adoption improves, software investments become more intentional, and innovation happens organically instead of through mandates.

Anthropic announced in June that its most advanced Mythos-class models, including Fable, would be excluded from Zero Data Retention (ZDR) agreements. Each organization will need to conduct a risk versus benefit analysis to decide how best to approach the use of these “covered models.” While the same risk of review and retention impacts us all, some types of legal work will likely trigger more reviews than others.