Build an AI judge for small disputes.
When small disputes recur, have the people involved work through a shared way to examine evidence and use AI recommendations. Give each side a serious advocate and a third AI to weigh the cases. People retain their own decision rights; an AI judge is not automatically impartial, binding or accepted.
Try it now
Copy the prompt into your AI tool of choice, then answer its questions as they come.
Paste it into any AI chat to start. For most people, this is all you need.
Everything on this page in one paste: the prompt plus the full plan, every step and what to check. Use it for longer moves, to plan ahead, to share with a colleague, or if your AI can't open links.
How it goes
What the AI will do
- Investigate the dispute, evidence and existing rules; arrange independent advocacy and an explained recommendation.
- Build and test a reusable workflow and revise it as actual participants work through the recurring procedure.
What to have ready
- A recurring class of disputes, relevant evidence and the actual people involved.
- Their actual accounts and contributions to the shared process through an agreed channel.
Where you come in
- Have participants confirm their own accounts and work through evidence access, corrections and use of recommendations.
- Review the real case and reusable procedure without treating an AI result as automatically neutral or binding.
What you'll get
A tested judging workflow, an explained real-case recommendation and a participant-worked-through procedure for recurring disputes, with decision rights and unresolved agreement clear.
You're done when
The people involved have worked through a shared procedure for recurring disputes, including evidence contributions, corrections and how judgments inform their decisions, and have tried the reusable judging tool on a real case.
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result. ### Start with the activity and the user's context Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear. Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them. ### Check the capabilities needed for this work Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit. When useful access is missing: - Explain what you cannot reach and how that material would help. - Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot. - If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response. - Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available. ### Ask focused questions and respect the answers Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful. Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements. ### Keep evidence and independent contributions distinct Distinguish evidence, inference, and missing information. Check your work against the source material. Use independent agents when your system actually supports delegation and independent investigation or review would improve the work: - Give each agent a bounded brief and the context appropriate to its role. - Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations. - Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately. ### Produce, check, and revise the result Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them. Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result. Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized. Help me build and use an AI-supported shared procedure for recurring disputes. Investigate the situation, arrange a serious hearing of the different positions, produce an explained judgment and make the process reusable. The intended result is a shared procedure for recurring disputes worked through by the actual people involved. Private thinking can prepare it but does not complete this version. Do not assume that an AI judgment is neutral, binding or accepted by the participants. ### Establish the use and find the record Start with the dispute or dilemma I supplied. Establish the recurring class of disputes and actual people who would use the shared procedure. Ask one focused question if that scope is unclear and wait. A private dilemma or a single exploratory hearing can inform preparation but does not by itself establish this shared-work arrangement; do not invent participants. Inspect and try relevant access. Search emails, chats including WhatsApp where accessible, calendar-linked discussions, meeting transcripts, agreements, policies, project requests and decision records. Follow the issue, people, dates and disputed terms. Establish what happened, what each person actually requested, previous attempts to resolve it, operative rules and the practical decision at stake. Trace later amendments to the earlier agreement rather than assuming the latest mention wins. Identify participants, their roles and stated concerns from actual correspondence and records. Keep their own words, my account of their position and your inference separate. If access is missing, investigate supported connection or export routes and ask for the specific record needed, explaining what it would establish. An interview or pasted account is useful but cannot recover unseen evidence or establish the other party's agreement. ### Agree a common hearing brief Produce a short brief: the question to decide, positions or competing considerations, agreed facts, disputed facts, relevant rules and desired outcome. Distinguish disputes about evidence from differences in values or priorities. Do not force a binary when a settlement or third option may satisfy the underlying concerns. For a shared hearing, give each participant the chance to confirm or correct their own account and supply relevant evidence through the agreed channel. Prepare the invitations or contribution forms yourself; sending requires authorization. Establish which material all participants and the adjudicator may see. Do not quietly brief one side with confidential material that cannot fairly enter the hearing. Missing participation means an incomplete hearing, not consent. Agree how the judgment will be used: exploratory advice, a proposed settlement or a recommendation the people will decide whether to accept. Keep this tool as decision support; it cannot acquire legal or organizational authority by calling itself an arbitrator. In a private exercise, use clearly labeled simulated positions. Do not contact or speak for another person. Confirm consequential gaps in the decision criteria before adjudication. If no rule settles a tradeoff, show the values choice rather than inventing a company rule. Preserve these choices in the reusable brief. ### Give each side a separate advocate Where available, create one independent agent for each materially different side or consideration. Give each the same common facts, rules and admissible evidence with source references, plus the position it must examine. Withhold the other advocates' first arguments and your preferred verdict. Ask each to produce the strongest honest case, cite its support, acknowledge contrary evidence, identify what it wants and name missing facts that could change its position. Advocacy must not fabricate evidence. Size the evidence packets to what an agent can inspect. For a large record, use bounded source investigations before the hearing, preserving full conversational context and a coverage record. Use successive waves if capacity is limited. If independent agents are unavailable, prepare fresh-session briefs with an explicit return path. If only sequential role passes are possible, explain that they share context and let me choose that limited format before presenting it as a substitute. After independent first cases are complete, allow a bounded response to the actual opposing case. Ask each advocate to identify the pivotal disagreement and a possible resolution, not merely repeat its opening. Preserve original cases and responses. ### Have a separate adjudicator decide Give a fresh adjudicator the common brief, source evidence, final cases and responses, without your preferred outcome or the advocates' private drafting commentary. It should verify pivotal claims, apply the agreed criteria consistently and explain which arguments survive examination. Agent agreement is not proof; resolve factual disagreements against the record. Require a plain-language judgment containing the question, recommended resolution, decisive evidence, treatment of each side's strongest point, unresolved facts and what would change the judgment. If evidence or criteria cannot support a winner, it may recommend a conditional settlement or decline to decide until a named gap is filled. It should not award certainty merely because the requested format says judge. Show me the judgment, and in a shared process make it available through the authorized common channel. Invite corrections of material facts or omitted arguments. Obtain additional accessible evidence and rerun the affected adjudication when warranted. Record why the outcome changed, or why it did not. A person's acceptance remains their actual response, never an inference from silence. ### Build and test the reusable tool Use the available permitted AI workspace, reusable prompt or small interface to retain the hearing brief, source requirements, advocate and adjudicator instructions, output format and correction route. Make each new case use its own evidence; do not let a previous case's outcome become a hidden rule. Give participants a simple way to start a case through the shared procedure. Label any private preparation separately. Try the tool on the actual case and, where available, a past case or a deliberately incomplete example. Check that both sides' evidence reaches the adjudicator, important citations support the judgment, missing information remains visible and correcting a pivotal fact triggers reconsideration. A historical human outcome is a comparison, not an automatic correctness label. Inspect the saved artifact and demonstrate how I can use it again. If the intended shared interface cannot be built, deliver a runnable prompt workflow and state the unbuilt sharing component; do not call a mock interface operational. ### Work through the recurring procedure with the actual participants Use the hearing and tested workflow to help people establish how recurring disputes will be handled. Prepare concrete choices about who contributes evidence, what each participant can inspect, how accounts are corrected and how recommendations inform the people's own decisions. Do not introduce binding legal or organizational authority or remove existing decision rights without an explicit authorized decision. Bring the actual participants into working through this procedure through a discussion I host or an agreed document, and use their actual notes, transcript or contributions. Do not send invitations without authorization or substitute AI advocates for human participation. Record what changes in their shared way of handling disputes, actual responses and material disagreement. Acceptance of a procedure is distinct from accepting a particular AI judgment; neither follows from silence. Revise the reusable brief and workflow from that joint work and recheck affected evidence/correction paths. This can finish at joint preparation for the recurring process; a live organizational rollout is not required. Private thought or one shared hearing alone remains preparation, not completion of the new shared practice. Finish with the reusable tool, case brief and arguments, explained judgment and the recurring procedure worked through by the actual participants. Identify what changes in their practice and any unresolved agreement. If only private preparation or a single hearing has occurred, state that the shared-work activity remains unfinished. Leave substantive decisions and external actions with the people involved. These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
View full AI instructions
Build an AI judge for small disputes.
What the AI will do
- Investigate the dispute, evidence and existing rules; arrange independent advocacy and an explained recommendation.
- Build and test a reusable workflow and revise it as actual participants work through the recurring procedure.
What you can provide
- A recurring class of disputes, relevant evidence and the actual people involved.
- Their actual accounts and contributions to the shared process through an agreed channel.
Where you'll take part
- Have participants confirm their own accounts and work through evidence access, corrections and use of recommendations.
- Review the real case and reusable procedure without treating an AI result as automatically neutral or binding.
What you'll get
A tested judging workflow, an explained real-case recommendation and a participant-worked-through procedure for recurring disputes, with decision rights and unresolved agreement clear.
Activity context
When small disputes recur, have the people involved work through a shared way to examine evidence and use AI recommendations. Give each side a serious advocate and a third AI to weigh the cases. People retain their own decision rights; an AI judge is not automatically impartial, binding or accepted.
For the user
Bring a recurring class of disputes and relevant records. Have the actual people involved confirm their accounts and work through how evidence, corrections and recommendations will be handled. AI builds and tries the judging workflow and revises it from their actual contributions. Private thought or one hearing alone does not complete this version.
Guidance for the AI
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.
Start with the activity and the user's context
Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.
Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.
Check the capabilities needed for this work
Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.
When useful access is missing:
- Explain what you cannot reach and how that material would help.
- Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
- If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
- Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.
Ask focused questions and respect the answers
Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.
Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.
Keep evidence and independent contributions distinct
Distinguish evidence, inference, and missing information. Check your work against the source material.
Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:
- Give each agent a bounded brief and the context appropriate to its role.
- Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
- Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.
Produce, check, and revise the result
Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.
Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.
Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.
Your AI prompt
Help me build and use an AI-supported shared procedure for recurring disputes. Investigate the situation, arrange a serious hearing of the different positions, produce an explained judgment and make the process reusable. The intended result is a shared procedure for recurring disputes worked through by the actual people involved. Private thinking can prepare it but does not complete this version. Do not assume that an AI judgment is neutral, binding or accepted by the participants.
Establish the use and find the record
Start with the dispute or dilemma I supplied. Establish the recurring class of disputes and actual people who would use the shared procedure. Ask one focused question if that scope is unclear and wait. A private dilemma or a single exploratory hearing can inform preparation but does not by itself establish this shared-work arrangement; do not invent participants.
Inspect and try relevant access. Search emails, chats including WhatsApp where accessible, calendar-linked discussions, meeting transcripts, agreements, policies, project requests and decision records. Follow the issue, people, dates and disputed terms. Establish what happened, what each person actually requested, previous attempts to resolve it, operative rules and the practical decision at stake. Trace later amendments to the earlier agreement rather than assuming the latest mention wins.
Identify participants, their roles and stated concerns from actual correspondence and records. Keep their own words, my account of their position and your inference separate. If access is missing, investigate supported connection or export routes and ask for the specific record needed, explaining what it would establish. An interview or pasted account is useful but cannot recover unseen evidence or establish the other party's agreement.
Agree a common hearing brief
Produce a short brief: the question to decide, positions or competing considerations, agreed facts, disputed facts, relevant rules and desired outcome. Distinguish disputes about evidence from differences in values or priorities. Do not force a binary when a settlement or third option may satisfy the underlying concerns.
For a shared hearing, give each participant the chance to confirm or correct their own account and supply relevant evidence through the agreed channel. Prepare the invitations or contribution forms yourself; sending requires authorization. Establish which material all participants and the adjudicator may see. Do not quietly brief one side with confidential material that cannot fairly enter the hearing. Missing participation means an incomplete hearing, not consent.
Agree how the judgment will be used: exploratory advice, a proposed settlement or a recommendation the people will decide whether to accept. Keep this tool as decision support; it cannot acquire legal or organizational authority by calling itself an arbitrator. In a private exercise, use clearly labeled simulated positions. Do not contact or speak for another person.
Confirm consequential gaps in the decision criteria before adjudication. If no rule settles a tradeoff, show the values choice rather than inventing a company rule. Preserve these choices in the reusable brief.
Give each side a separate advocate
Where available, create one independent agent for each materially different side or consideration. Give each the same common facts, rules and admissible evidence with source references, plus the position it must examine. Withhold the other advocates' first arguments and your preferred verdict. Ask each to produce the strongest honest case, cite its support, acknowledge contrary evidence, identify what it wants and name missing facts that could change its position. Advocacy must not fabricate evidence.
Size the evidence packets to what an agent can inspect. For a large record, use bounded source investigations before the hearing, preserving full conversational context and a coverage record. Use successive waves if capacity is limited. If independent agents are unavailable, prepare fresh-session briefs with an explicit return path. If only sequential role passes are possible, explain that they share context and let me choose that limited format before presenting it as a substitute.
After independent first cases are complete, allow a bounded response to the actual opposing case. Ask each advocate to identify the pivotal disagreement and a possible resolution, not merely repeat its opening. Preserve original cases and responses.
Have a separate adjudicator decide
Give a fresh adjudicator the common brief, source evidence, final cases and responses, without your preferred outcome or the advocates' private drafting commentary. It should verify pivotal claims, apply the agreed criteria consistently and explain which arguments survive examination. Agent agreement is not proof; resolve factual disagreements against the record.
Require a plain-language judgment containing the question, recommended resolution, decisive evidence, treatment of each side's strongest point, unresolved facts and what would change the judgment. If evidence or criteria cannot support a winner, it may recommend a conditional settlement or decline to decide until a named gap is filled. It should not award certainty merely because the requested format says judge.
Show me the judgment, and in a shared process make it available through the authorized common channel. Invite corrections of material facts or omitted arguments. Obtain additional accessible evidence and rerun the affected adjudication when warranted. Record why the outcome changed, or why it did not. A person's acceptance remains their actual response, never an inference from silence.
Build and test the reusable tool
Use the available permitted AI workspace, reusable prompt or small interface to retain the hearing brief, source requirements, advocate and adjudicator instructions, output format and correction route. Make each new case use its own evidence; do not let a previous case's outcome become a hidden rule. Give participants a simple way to start a case through the shared procedure. Label any private preparation separately.
Try the tool on the actual case and, where available, a past case or a deliberately incomplete example. Check that both sides' evidence reaches the adjudicator, important citations support the judgment, missing information remains visible and correcting a pivotal fact triggers reconsideration. A historical human outcome is a comparison, not an automatic correctness label. Inspect the saved artifact and demonstrate how I can use it again. If the intended shared interface cannot be built, deliver a runnable prompt workflow and state the unbuilt sharing component; do not call a mock interface operational.
Work through the recurring procedure with the actual participants
Use the hearing and tested workflow to help people establish how recurring disputes will be handled. Prepare concrete choices about who contributes evidence, what each participant can inspect, how accounts are corrected and how recommendations inform the people's own decisions. Do not introduce binding legal or organizational authority or remove existing decision rights without an explicit authorized decision.
Bring the actual participants into working through this procedure through a discussion I host or an agreed document, and use their actual notes, transcript or contributions. Do not send invitations without authorization or substitute AI advocates for human participation. Record what changes in their shared way of handling disputes, actual responses and material disagreement. Acceptance of a procedure is distinct from accepting a particular AI judgment; neither follows from silence.
Revise the reusable brief and workflow from that joint work and recheck affected evidence/correction paths. This can finish at joint preparation for the recurring process; a live organizational rollout is not required. Private thought or one shared hearing alone remains preparation, not completion of the new shared practice.
Finish with the reusable tool, case brief and arguments, explained judgment and the recurring procedure worked through by the actual participants. Identify what changes in their practice and any unresolved agreement. If only private preparation or a single hearing has occurred, state that the shared-work activity remains unfinished. Leave substantive decisions and external actions with the people involved.
Done when
The people involved have worked through a shared procedure for recurring disputes, including evidence contributions, corrections and how judgments inform their decisions, and have tried the reusable judging tool on a real case.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.