Wrap up

Run the Legal AI Research Loop

Put the course together: prompt, roadmap, source-backed authority, follow-up questions, and verification.

You will leave with a repeatable loop for using Claude, ChatGPT, or Perplexity as a supervised legal research assistant.

What you can do now

You now have the core research habit: give the assistant safe context, ask for a roadmap, make it use legal source data, then keep asking follow-up questions until the source trail is visible.

That is the foundation for better drafting, reusable workflows, and more advanced AI work later. This course stops at the most important beginner skill: getting to source-backed research you can inspect.

The loop

  1. 1Start with a tool and account that are appropriate for the information you plan to use.
  2. 2Frame the question with jurisdiction, issue, posture, safe facts, output, and verification expectations.
  3. 3Ask for a roadmap before a conclusion so you can correct the plan early.
  4. 4Use Descrybe Legal Engine when the answer needs cases, statutes, rules, citations, source text, quoted language, or treatment signals.
  5. 5Ask follow-up questions about any case, citation, quote, treatment issue, or proposition that matters.
  6. 6Check source support before relying on the result.
  7. 7Write down what is confirmed, what is uncertain, and what still needs human legal judgment.

Capstone prompt

Try this with a low-risk public, hypothetical, or redacted issue. The goal is to practice the loop, not to get a finished legal answer in one pass.

Research loop prompt
I want to practice a source-backed legal research loop. Issue: [issue]. Jurisdiction: [jurisdiction]. Posture or task: [posture or task]. Facts: [safe public, hypothetical, or redacted facts]. First build a research roadmap before giving any conclusion. Then use Descrybe Legal Engine to find a small starting set of relevant authority. For each authority, explain why it matters and what needs checking. After that, suggest targeted follow-up questions I should ask before relying on the research.

Keep these follow-ups handy

  • Why did you include this authority?
  • What proposition does it actually support?
  • What facts, posture, or jurisdiction limits matter?
  • Is this still good law for the point I need?
  • What authority cuts the other way?
  • Show me the source text for that point.
  • What should I check next before relying on this?

Where to go next

Drafting from verified research and saving reusable workflows are natural next topics, but they need more room, more examples, and more supervision than this starter course should try to squeeze in.

For now, if you can frame a research question, ask for a roadmap, find authority, follow up on results, and check support against sources, you are already using AI in a much more careful way than a one-shot prompt allows.