Legal AI has learned to summarise. Has it learned how lawyers think?

Most legal AI tools make summarising, retrieving and drafting faster, whereas faster summaries do not necessarily produce better legal reasoning, and AI can still misread context or miss what matters. In my experience the real craft of law lies in judgment, i.e. reading, identifying what matters, spotting contradictions and deciding how the facts shape an argument, and one basic way lawyers exercise that judgment is that they read and they highlight.

That is where "May or Shall" begins: you highlight the passages that matter, each keeps its source, and those selections become the foundation for drafting with ChatGPT. We launch here soon and I would value your view before we do.

When you use AI on a long document, what do you check first in what it gives back, and would starting from your own highlights change that?