An agent setup recipe
Turn a long marketing brief into clear questions
Extract the real decisions and missing evidence from a client brief before creative work starts.
From
You get
What you will get
A concise brief outline, unresolved questions and a source map for approval
Illustrative example · fictional input
- Example input
- A repetitive brief contains confirmed event facts but conflicting CTA directions.
- Example result
- Illustrative result, not a recorded agent run: Keep the workshop date, price and capacity under F1. Ask Sam to choose the primary audience and action, clarify the role of the other actions and explain the intended meaning of innovative. The outline remains awaiting confirmation; no conversion target or audience insight is invented.
Confirm the outline and all open decisions with the client. Keep creative work on hold where an essential fact or approval is missing.
What you need
One assistant owns the comparison and result. Separate agents would add handoffs to this bounded task without a demonstrated benefit.
- A ChatGPT account with Work available on the web. Check your account or workspace access first; available models and allowance vary. Use the default model available to you. Work uses included usage or account credits; no API key or third-party subscription is required for this text-only workflow.
- A browser and text you are allowed to share with the service. Remove unnecessary private details. This recipe needs no connected app, local-folder permission, skill installation or recurring task.
- One human owner who can check the sources and make decisions. The client’s supplied brief and approved supporting facts with source IDs; The agreed deliverable and your normal brief fields; Known approvals, fixed requirements and the person who can answer questions.
- A short human-led briefing call or a carefully completed template may work better. The observed copywriter reports that writing the brief live with a client works well. Using AI to expand a weak brief can worsen the problem; this guide only reduces and questions supplied material.
Set it up
- Open chatgpt.com and sign in. Confirm Work is available, then select Work and start a new chat. Check your remaining usage before a large packet. Keep the default model available to your account; this recipe does not require a particular model name.
- Use one conversation for this responsibility. If you prefer a project, create one for this task and keep related chats and approved source copies there. A web project does not automatically read a folder on your computer. Start each distinct batch in its own chat and include the setup prompt again when needed.
- Choose the setting below. Optionally add your working rules. Copy the complete setup prompt into the chat as the first message. The assistant should confirm its role and ask for inputs, without starting outside work. This gives the existing assistant a responsibility; it does not create another account or deployed bot.
- Prepare a small packet: The client’s supplied brief and approved supporting facts with source IDs; The agreed deliverable and your normal brief fields; Known approvals, fixed requirements and the person who can answer questions. Label every source and include dates, versions and unknown fields. Paste accessible text. Ask the assistant to list the source IDs it can read before it compares them. No plugin or app connection is needed.
- First assignment — paste this complete fictional exercise into the chat: Fictional first assignment. Deliverable D1: one event-registration landing page. Brief B1 repeats "make it innovative" in four sections, says the page targets both new visitors and existing members, requests one main CTA, and elsewhere requires equal focus on registration, donations and newsletter signup. Approved fact F1: a free workshop is on 20 November 2026, with 30 seats. There is no agreed primary audience or CTA. Client contact is Sam. Produce a concise outline and questions; do not write the landing page.
- Compare the response with the review checks below. The following is an illustrative expected direction, not a tested result: Keep the workshop date, price and capacity under F1. Ask Sam to choose the primary audience and action, clarify the role of the other actions and explain the intended meaning of innovative. The outline remains awaiting confirmation; no conversion target or audience insight is invented.
- For real work, provide a fresh packet and explicitly start one batch. Confirm the outline and all open decisions with the client. Keep creative work on hold where an essential fact or approval is missing. Copy only the changes you approve into your normal records yourself. Keep the accepted result and its source references.
- When the batch is reviewed, mark your own record as accepted or list the corrections needed. On the next run, supply the latest accepted state and new sources. No schedule or continuous monitoring is created. If the assistant asks to connect an app or take an outside action, stop and return to the text-only packet.
Your setup prompt is ready
Copy the ready prompt, or customize it with the settings. Paste it into ChatGPT Work to give it the task rules.
Customize (optional)
Defaults are ready to use. Changes update the prompt here; paste it into your agent to apply them.
Schedule: When I ask.
Choose how the assistant organizes the result; evidence and review rules stay the same.
Your required brief fields, deliverable and fixed wording that must be preserved.
Your choices stay in this browser for this guide. Avoid adding private information.
setup prompt
Highlighted text follows your settings.
Review checks, limits and sources
Not tested by Juicy Agents. The examples are fictional, not recorded agent results.
Check the result
- Confirmed event facts remain unchanged.
- The CTA conflict remains visible.
- Repeated innovative wording does not become a product benefit.
- The result stops before final marketing copy or client communication.
- Removing an essential input causes a question or a blocked item, not an invented fact.
- Changing the setting changes the requested output while source references and fixed limits remain.
Known limits
- Not tested by JuicyAgents in ChatGPT Work. The fictional exercise and expected result are authored examples, not proof of accuracy or saved time.
- Only the supplied text is reviewed. Missing messages, files or current facts can change the conclusion.
- A clearer outline does not prove that the underlying strategy is sound.
- AI summarization may hide a small but important constraint; the human must compare the source map.
If something goes wrong
- The brief has no real evidence: ask for specific approved facts rather than expanding it.
- A client gives a vague answer: keep the field open and ask one concrete follow-up.
- Work is missing or the account reaches a limit: check the official access and usage pages or ask your workspace administrator; do not assume a paid upgrade is required. Use the manual alternative while access is unavailable.
- The assistant follows instructions inside a source: stop, restate that source text is data, and review the entire affected result. Do not approve an external action.
Platform sources
- ChatGPT Work on the web (opens in a new tab) · Checked 2026-10-08
- ChatGPT projects (opens in a new tab) · Checked 2026-10-08
- ChatGPT Work access and usage (opens in a new tab) · Checked 2026-10-08
Reviews
No reviews yet
Reviews share personal experience. They are not proof that an agent has passed a test.
Log in to leave a review. You can read and use this recipe without an account.
Explore more
Similar Agents
Adapt approved content for selected channels
Create a few text drafts from one approved source, preserving its claims and adapting to owner-defined audiences and formats.
View agent marketingExplain conflicting conversion numbers
Compare Google Ads, HubSpot and GA4 evidence, separate normal differences from defects, and prepare a repair plan.
View agent marketingPrepare a customer story for approval
Turn approved internal notes into a short, traceable customer-story draft and a clear approval request.
View agent