Role-First Multi-Shot Prompting
Assign a role, define the task, add context, then refine the response
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 96%
Role-First Multi-Shot Prompting improves AI output through staged context and deliberate iteration. First, the user tells the model what kind of expert it should emulate, narrowing the perspective and capabilities relevant to the task. Next, the user specifies the deliverable, audience, tone, source material, and constraints. The initial response is treated as a working draft. Instead of restarting, the user identifies what worked or failed and issues focused follow-ups, such as expanding one point, shortening the copy, or changing its tone. For customer-facing work, an ideal-client persona can be developed first and retained in the conversation so subsequent copy addresses a concrete person rather than an undefined market.
Origin
Extracted from The Jasmine Star Show during Chris Winfield and Jen Gottlieb's explanation of how beginners can obtain stronger results from conversational AI.
Core principles
- 01Give the AI a specific expert identity before assigning work
- 02Describe the desired task and outcome precisely
- 03Supply audience, offer, voice, and tone context
- 04Treat the first response as a draft rather than a finished asset
- 05Refine promising elements through focused follow-up prompts
How to run it
- 1
Assign an expert identity
Tell the AI who it should be for this task, such as a conversion copywriter, customer researcher, or strategist. Choose a role directly connected to the desired result.
Pro tip Include the expert's relevant training or specialty instead of using a broad label.
Watch out Do not assume the model will infer the right perspective from the task alone.
- 2
Define the deliverable
State exactly what the model should create and where it will be used. Include practical constraints such as length, format, and call to action.
Pro tip Describe what success should look like for the reader or business.
Watch out A vague request such as “make this better” can produce superficial changes.
- 3
Supply audience and context
Provide the offer, customer profile, tone, examples, and any source material needed to complete the task. Build a detailed customer persona first when audience fit is central.
Pro tip Give the persona a name and concrete motivations so later prompts can refer to one recognizable customer.
Watch out Do not ask the model to invent essential business facts that you already possess.
- 4
Generate the first draft
Ask for an initial response and assess it as a draft. Identify useful sections as well as omissions, weak language, or tonal problems.
Pro tip Compare the result with the actual objective rather than judging it only for fluency.
Watch out Do not publish the first response simply because it sounds polished.
- 5
Refine with follow-ups
Continue the same conversation with specific revision instructions. Ask the model to deepen a strong point, remove weak material, shorten the response, or change its style.
Pro tip Reference the exact section you want preserved or expanded.
Watch out Repeatedly requesting a generic rewrite may erase the strongest parts of the draft.
- 6
Verify and personalize
Check facts, claims, brand voice, and audience relevance before using the output. Add personal judgment and original experience so the final asset belongs to the business.
Pro tip Use AI as a writing partner and creative catalyst rather than an autonomous publisher.
Watch out Never rely on stylistic confidence as evidence of factual accuracy.
In the wild
An entrepreneur first asks the AI to create a detailed persona for the company's ideal buyer. In the same conversation, the entrepreneur assigns the model the role of a conversion copywriter, supplies the offer's bullet points, requests a fun conversational tone, and asks for an ad directed specifically to the named persona. After reviewing the first draft, the entrepreneur asks the model to deepen its strongest benefit and shorten the opening.
→ The entrepreneur receives a more audience-specific draft that can be fact-checked and personalized before publication.
A founder struggling with a landing-page section pastes the unfinished copy into an AI conversation. The founder assigns the model an appropriate writing role, explains the product and customer, and requests several clearer approaches. One suggestion restores the founder's creative momentum, and the founder rewrites it in the brand's own language.
→ AI supplies creative direction without replacing the founder's voice or judgment.
Common mistakes
Skipping the role
Giving the task without first establishing a relevant perspective can result in a broad, subpar response that lacks expert focus.
Leaving out audience context
Copy aimed at an undefined public will usually be less relevant than copy grounded in a detailed ideal-client persona.
Accepting the first draft
The first response may contain useful raw material, but strong output often requires several precise follow-up instructions.
Is it for you?
Best for
It is best for entrepreneurs using conversational AI to draft copy, research customers, or develop content.
Not ideal for
It is not ideal for factual or regulated work that will be published without independent verification.
From the transcript
“And the first thing that you want to prompt it to do is who you want it to be.”
“And then the most important part that, like even Chris taught me, is your first outcome might not be that great.”
“I really liked point number two. Can you go a little bit deeper into point number two”
From the episode
The BEST AI Strategies to Grow Your Business with Chris Winfield
Chris Winfield