A reusable prompt that turns any AI into a senior partner who challenges your draft before the client sees it.
Click any node to expand. 10 structured challenges, then revise and repeat.
The AI asks YOU up to 5 questions before it starts critiquing. Context changes everything — don't let it guess. This forces you to surface information you assumed was obvious.
Every draft carries assumptions the writer doesn't notice. This names them and explains why a judge, regulator, or sophisticated client might not share them.
The single strongest argument against your position, written in opposing counsel's voice at full strength. If you can't answer it, your advice isn't ready.
Statutes, cases, regulations that should have been cited. Warning: AI models invent citations. Verify every authority independently.
Sentences a hostile reader could read two ways. Both readings shown. Tighter wording recommended. Ambiguity is where disputes are born.
Does this solve the client's real problem or just the technical one? Legally correct but commercially naive advice is still bad advice.
The iceberg. Risks, edge cases, and worst-case scenarios the draft doesn't mention. A good advice letter warns — what's absent?
Too confident where it should hedge? Too hedged where it should be direct? Clients pay for a view, not a menu of options.
The single most important issue to address before this goes out. Not a list. Not a top three. One issue. Forces prioritisation.
Context the AI discovered was absent only after doing the work. Reveals gaps in your thinking, not just your drafting.
The AI asks YOU up to 5 questions before it starts critiquing. Context changes everything — don't let it guess. This forces you to surface information you assumed was obvious.
Every draft carries assumptions the writer doesn't notice. This names them and explains why a judge, regulator, or sophisticated client might not share them.
The single strongest argument against your position, written in opposing counsel's voice at full strength. If you can't answer it, your advice isn't ready.
Statutes, cases, regulations that should have been cited. Warning: AI models invent citations. Verify every authority independently.
Sentences a hostile reader could read two ways. Both readings shown. Tighter wording recommended. Ambiguity is where disputes are born.
Does this solve the client's real problem or just the technical one? Legally correct but commercially naive advice is still bad advice.
The iceberg. Risks, edge cases, and worst-case scenarios the draft doesn't mention. A good advice letter warns — what's absent?
Too confident where it should hedge? Too hedged where it should be direct? Clients pay for a view, not a menu of options.
The single most important issue to address before this goes out. Not a list. Not a top three. One issue. Forces prioritisation.
Context the AI discovered was absent only after doing the work. Reveals gaps in your thinking, not just your drafting.
The core judgement comes from the lawyer. AI compresses turnaround and catches the gap at 11pm. It doesn't originate the thinking.
Quality thinking used to depend on who was in the room. AI gives every lawyer a sparring partner that won't flatter. Available at 10pm.
Don't accept the first output. Run structured critique. Fix what's weak. Run it again. The second pass surfaces a different layer.
Copy it. Paste into any AI tool. Paste your draft underneath.
Save the prompt in your platform so you never paste it again.
Projects → New Project → paste into Project instructions. Upload your firm's style guide for context.
Explore GPTs → Create → paste into Instructions. Keep Private. Paste draft, send.
Prompt library → Save a prompt. Pin it. M365 commercial keeps data in tenant.
Gems → New Gem → paste prompt. Save. Workspace Gems go firm-wide.
Do not paste client-identifying information into any AI tool not approved by your firm. In many Australian states, this may breach the Australian Solicitors' Conduct Rules. Sanitise all drafts. No client names. No matter numbers.
NewLaw helps Australian law firms select, implement, and govern AI tools — from readiness assessments to full rollouts.