Split a project between many AI assistants.
Several assistants can make a big project manageable—when one lead makes their work fit together.
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 your project’s goal, existing work and decisions, then divide suitable work into bounded assignments.
- Coordinate the assistants, assemble the finished result, and use a separate reviewer where available to find and repair gaps.
What to have ready
- A project, rough brief or starting files, with access to relevant documents, emails, chats, meeting records and deadlines.
- Any known requirements for the result and the people who will use it.
Where you come in
- Confirm consequential scope choices and resolve questions the records cannot answer.
- If agents are unavailable, choose whether to relay prepared handoffs between fresh sessions or use one AI in manageable passes; review the finished result.
What you'll get
The agreed project deliverable, assembled and checked, with any unfinished work explicit; a division-of-work plan alone is not completion.
You're done when
You have the agreed usable project deliverable and evidence that it meets the important requirements. Any missing access, undecided scope or unfinished work is clearly identified rather than presented as a completed project.
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 complete a real project by dividing suitable work among AI assistants. Own the coordination, integration and corrections so that I receive the agreed project result, not merely a list of agent assignments. ### 1. Establish the project and what finished means Start with the project or rough material I provide. Inspect the brief, current project files and task records; search relevant emails, chats and other messages, meeting transcripts, decisions, calendar milestones and deadlines. Use project names, participants and terms from the initial material to find existing work, agreements, dependencies and changes of direction. Determine my role, intended audience, practical use and what is already complete. Check dates and decision authority rather than assuming the latest file timestamp settles the goal. Summarize the intended deliverable, the work remaining and observable acceptance criteria. For example, a market assessment needs supported findings and a coherent recommendation; a working tool needs implemented behavior and a successful test of its main use. These are examples, not mandatory project types. Ask focused questions one at a time when scope, audience or a consequential choice is unclear. Confirm a scope you infer on my behalf before dependent production begins; an already explicit brief can be sufficient. Continue independent investigation while awaiting answers. ### 2. Check capacity and choose the division of work Inspect the available agent, file, browsing and production tools. Try the relevant source access before asking me to assemble material. If access is missing, investigate connections or exports, prepare what can be prepared and guide the specific action needed. Identify which deliverables are blocked by absent material rather than claiming you can produce them from recollection. Inspect whether you can actually start agents, give them isolated contexts, pass files or source references, receive results and run reviews. Decide which work benefits from division. Keep closely coupled reasoning or a single narrative under one lead; separate genuinely independent research areas, artifacts, specialist tasks or reviews. Do not create agents merely to meet a number. Map dependencies and size each assignment to what one agent can inspect reliably. Avoid breaking a conversation or dataset into pieces that lose necessary context. Plan successive waves if concurrency is limited. Keep a coverage record listing the relevant source sets, assigned units, completed work, unresolved gaps and integration dependencies. Include overlaps only when they serve comparison or verification. If autonomous agents are unavailable, offer a workable division into fresh sessions with copyable handoffs and an explicit return path. Explain any relay I must perform and confirm that route before relying on it. Prepare the handoffs yourself; do not make me invent briefs or integrate the outputs. If I prefer one session, complete manageable passes and identify that as sequential work. A lack of fresh contexts means the contributions are not independent. Narrowing the agreed project because of tool limits requires my decision; it is not an invisible change in the promised result. ### 3. Brief and run the work Give each worker a bounded brief containing: - The project goal, its role and the exact question or artifact it owns. - Relevant source material or reachable references, necessary definitions and prior decisions. - What it may change, what it must leave to other workers, and any dependency that must finish first. - A concrete output format, citation or test evidence, and criteria for a complete assignment. - Unresolved questions to return rather than guess, and where to place its output. For independent discovery or alternative perspectives, keep early findings separate until the first passes finish. Give audience workers records establishing the actual people, their expressed concerns and the material they will encounter; label any simulated response. Assign separate output files or clearly separated editing areas so workers do not overwrite one another. Ask workers to report inspected sources and unexamined coverage, not just a confident conclusion. Start ready assignments, collect their outputs, and check each against its brief. Return incomplete work for repair or reassign a bounded remainder. Resolve conflicting evidence by inspecting the sources, not by counting agent votes. Record the resolution or ask me about a consequential unresolved choice. Only then start work that depends on it. Preserve completed work and explicit handoffs if the project must continue across sessions; do not claim background continuation unless it is actually arranged. ### 4. Assemble the actual deliverable Integrate the contributions into the promised artifact. Reconcile terminology, dates, assumptions, source versions and repeated material. Write the connecting argument or implement the interfaces that turn separate pieces into one usable whole. Verify that the assembly covers every agreed requirement and that individually correct pieces remain consistent together. Run checks appropriate to the artifact: trace substantive claims to their sources, recalculate important figures, open the actual files, or exercise the tool's main interaction with representative inputs. Do the required production yourself or assign the remaining concrete work. A coordination plan, a stack of incompatible drafts or unexecuted build instructions is an intermediate result. ### 5. Review independently, repair and deliver Give a separate reviewer the actual assembled artifact, the agreed requirements and necessary source or test evidence, without the authors' explanation of why it should pass. Have it first assess what a user would understand or be able to do, then test fidelity and missing coverage. Split review into bounded sections when necessary, retaining a final integration review. Use a fresh-session handoff if possible when no reviewer agent is available; otherwise perform and label a dedicated shared-context review rather than claiming independence. Resolve material findings, update the artifact and rerun the affected checks. Recheck interactions when a local repair changes another part. Deliver the usable result with direct file links or access instructions, concise verification evidence and an accurate account of any remaining limitation. Invite my review of the actual result and incorporate corrections within the agreed project. Completion means the agreed work is delivered and checked. If essential access or a decision is still missing, hand over the completed portion and exact blocker as an unfinished result. Stop at the project boundary; publication or actions represented as my decisions follow the authorization already established for this project. 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
Split a project between many AI assistants.
What the AI will do
- Investigate your project’s goal, existing work and decisions, then divide suitable work into bounded assignments.
- Coordinate the assistants, assemble the finished result, and use a separate reviewer where available to find and repair gaps.
What you can provide
- A project, rough brief or starting files, with access to relevant documents, emails, chats, meeting records and deadlines.
- Any known requirements for the result and the people who will use it.
Where you'll take part
- Confirm consequential scope choices and resolve questions the records cannot answer.
- If agents are unavailable, choose whether to relay prepared handoffs between fresh sessions or use one AI in manageable passes; review the finished result.
What you'll get
The agreed project deliverable, assembled and checked, with any unfinished work explicit; a division-of-work plan alone is not completion.
Activity context
Several assistants can make a big project manageable—when one lead makes their work fit together.
For the user
Bring a project you want finished, even if the brief is rough. The AI will investigate the existing work, establish what success means and carry the project through assignments, integration and review. You will supply access, make consequential choices and review the result. If your tools cannot run agents, the AI will explain the workable session alternatives and any relay you would need to do.
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 complete a real project by dividing suitable work among AI assistants. Own the coordination, integration and corrections so that I receive the agreed project result, not merely a list of agent assignments.
1. Establish the project and what finished means
Start with the project or rough material I provide. Inspect the brief, current project files and task records; search relevant emails, chats and other messages, meeting transcripts, decisions, calendar milestones and deadlines. Use project names, participants and terms from the initial material to find existing work, agreements, dependencies and changes of direction. Determine my role, intended audience, practical use and what is already complete. Check dates and decision authority rather than assuming the latest file timestamp settles the goal.
Summarize the intended deliverable, the work remaining and observable acceptance criteria. For example, a market assessment needs supported findings and a coherent recommendation; a working tool needs implemented behavior and a successful test of its main use. These are examples, not mandatory project types. Ask focused questions one at a time when scope, audience or a consequential choice is unclear. Confirm a scope you infer on my behalf before dependent production begins; an already explicit brief can be sufficient. Continue independent investigation while awaiting answers.
2. Check capacity and choose the division of work
Inspect the available agent, file, browsing and production tools. Try the relevant source access before asking me to assemble material. If access is missing, investigate connections or exports, prepare what can be prepared and guide the specific action needed. Identify which deliverables are blocked by absent material rather than claiming you can produce them from recollection.
Inspect whether you can actually start agents, give them isolated contexts, pass files or source references, receive results and run reviews. Decide which work benefits from division. Keep closely coupled reasoning or a single narrative under one lead; separate genuinely independent research areas, artifacts, specialist tasks or reviews. Do not create agents merely to meet a number.
Map dependencies and size each assignment to what one agent can inspect reliably. Avoid breaking a conversation or dataset into pieces that lose necessary context. Plan successive waves if concurrency is limited. Keep a coverage record listing the relevant source sets, assigned units, completed work, unresolved gaps and integration dependencies. Include overlaps only when they serve comparison or verification.
If autonomous agents are unavailable, offer a workable division into fresh sessions with copyable handoffs and an explicit return path. Explain any relay I must perform and confirm that route before relying on it. Prepare the handoffs yourself; do not make me invent briefs or integrate the outputs. If I prefer one session, complete manageable passes and identify that as sequential work. A lack of fresh contexts means the contributions are not independent. Narrowing the agreed project because of tool limits requires my decision; it is not an invisible change in the promised result.
3. Brief and run the work
Give each worker a bounded brief containing:
- The project goal, its role and the exact question or artifact it owns.
- Relevant source material or reachable references, necessary definitions and prior decisions.
- What it may change, what it must leave to other workers, and any dependency that must finish first.
- A concrete output format, citation or test evidence, and criteria for a complete assignment.
- Unresolved questions to return rather than guess, and where to place its output.
For independent discovery or alternative perspectives, keep early findings separate until the first passes finish. Give audience workers records establishing the actual people, their expressed concerns and the material they will encounter; label any simulated response. Assign separate output files or clearly separated editing areas so workers do not overwrite one another. Ask workers to report inspected sources and unexamined coverage, not just a confident conclusion.
Start ready assignments, collect their outputs, and check each against its brief. Return incomplete work for repair or reassign a bounded remainder. Resolve conflicting evidence by inspecting the sources, not by counting agent votes. Record the resolution or ask me about a consequential unresolved choice. Only then start work that depends on it. Preserve completed work and explicit handoffs if the project must continue across sessions; do not claim background continuation unless it is actually arranged.
4. Assemble the actual deliverable
Integrate the contributions into the promised artifact. Reconcile terminology, dates, assumptions, source versions and repeated material. Write the connecting argument or implement the interfaces that turn separate pieces into one usable whole. Verify that the assembly covers every agreed requirement and that individually correct pieces remain consistent together.
Run checks appropriate to the artifact: trace substantive claims to their sources, recalculate important figures, open the actual files, or exercise the tool's main interaction with representative inputs. Do the required production yourself or assign the remaining concrete work. A coordination plan, a stack of incompatible drafts or unexecuted build instructions is an intermediate result.
5. Review independently, repair and deliver
Give a separate reviewer the actual assembled artifact, the agreed requirements and necessary source or test evidence, without the authors' explanation of why it should pass. Have it first assess what a user would understand or be able to do, then test fidelity and missing coverage. Split review into bounded sections when necessary, retaining a final integration review. Use a fresh-session handoff if possible when no reviewer agent is available; otherwise perform and label a dedicated shared-context review rather than claiming independence.
Resolve material findings, update the artifact and rerun the affected checks. Recheck interactions when a local repair changes another part. Deliver the usable result with direct file links or access instructions, concise verification evidence and an accurate account of any remaining limitation. Invite my review of the actual result and incorporate corrections within the agreed project. Completion means the agreed work is delivered and checked. If essential access or a decision is still missing, hand over the completed portion and exact blocker as an unfinished result. Stop at the project boundary; publication or actions represented as my decisions follow the authorization already established for this project.
Done when
You have the agreed usable project deliverable and evidence that it meets the important requirements. Any missing access, undecided scope or unfinished work is clearly identified rather than presented as a completed project.
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.