The real problem: a lack of method, not tools
The professionals coached by Revenu IA already use ChatGPT, Copilot, or an internal GPT. Yet they feel they are far from getting the most out of them. In many companies, the official tool is Microsoft Copilot, chosen for its security; in practice, users bypass it for other tools they deem more powerful. This is one of the sources of shadow AI.
Mustapha Benkalfate does not deny the differences between tools: Copilot can be less effective than Claude for certain uses, and so can ChatGPT. But in the vast majority of cases, the gap is less about raw power than about how the tool is used. With a solid method, quality results can be achieved regardless of the tool available in the company. SCOPE© is therefore not a trick: it's a principle, short for a simple task, more detailed for a sensitive one.
S: Specialist, narrowing the scope to increase relevance
An AI has been trained on billions of documents. Without an assigned role, it can draw on knowledge that is too broad and sometimes absurd for the need, "like cooking recipes or how to treat a whitlow," when a business analysis is expected. The specialist role sets the playing field, reduces noise, and directs the tone.
“You are an expert” is not enough. You need to specify: an expert in what, with what experience, in which sector, for what deliverable. For example: “You are an experienced business analyst with fifteen years of experience who is an expert in the IT services and consulting sector.”
C: Context, better to show than to explain
Context can be provided in two ways. Through text: the situation, constraints, the scope of the mission. And through documents: screenshots, advertisements, internal, regulatory, commercial, or HR documents. The golden rule of the method: with AIs, it's better to show than to explain. A significant part of quality loss comes from implicit information that the user thinks they have conveyed but the AI never received.
Practical tip: some tools consume a lot of tokens with PDFs. First converting a PDF to plain text in an initial conversation lightens the exchange and prevents getting stuck.
O: Objective, the task and the goal behind the task
Many users believe they have provided the objective just by stating the task. SCOPE© distinguishes two levels. The immediate objective: reply to an email, analyze a document. The higher objective: why. Is the goal of replying to this email to smooth over the relationship, reassure, reset expectations without being confrontational, or reopen a sales discussion? When the AI knows the intent, it no longer produces just a correct text, but a text geared towards the desired effect. Until you have answered "why do I really want to do this task?", your prompt is not finished.
P: Public, the most underutilized variable
AIs can finely adapt the register, tone, density, and arguments to an audience, provided it is described precisely. “Three Japanese investors in their forties” or “a group of scientific experts including two women aged 27, a man aged 24, and two seniors aged 53 and 57” produce very different texts. You can also provide the exchange history, the observed communication style, or CRM notes. The audience is not a detail: it is a design parameter.
E: Execution, staying in control and making the work auditable
Execution is where you regain control. Here, you set the tone, length, structure, and format of the deliverable. Above all, you require auditability: asking the AI to explain step-by-step the reasoning that led to its content, the data it used, and the patterns it deduced from the provided documents. This is what allows for verification, justification, and transmission. If the task is long or complex, it is broken down into several outputs rather than demanding a monolithic block: this is the purpose of the STEP© method.
The SCOPE© Prompt Template
S → Act as a [Specialist] in [Field]
C → Here is my situation: [Clear context, specific data]
O → My objective is to [what you want to accomplish]
P → The result will be read by [Public, profile, expectations]
E → Produce [Format] because [Reason], which [Action].
Tone: [Writing style].
SCOPE©: a proprietary Revenu IA method, © Mustapha Benkalfate. Reproduction of the prompt template is authorized with source attribution.
A real example, built live
During his July 2026 keynote in Kuala Lumpur, Mustapha Benkalfate drafted this type of prompt live in under two minutes: an AI literacy specialist who knows how to design interactive keynotes; the context of an 80-minute keynote at a time when AI regulations are a concern for boards of directors; the objective of addressing the risks of generative AI without being too somber; an audience of SME executives from Hong Kong; and a format starting with a summary, followed by a detailed description. The response independently drew on the European AI Act and a risk map.
Where does SCOPE© come from?
The method extends the “four pillars of prompting” described in his book AI Generation: role, context, audience, and output format. SCOPE© adds the distinction between the task and the higher objective, and the requirement of auditability.
When to use SCOPE© over another method
| Your situation | Method | What it does |
|---|---|---|
| You know exactly what you want | SCOPE© | Frames the request along five axes |
| Your need is still unclear | PROBE© | The AI probes, refines the need, checks the objective, briefs for nuances, then executes |
| The task is too complex for a single request | STEP© | Breaks it down into blocks, tasks, steps, processes |
| You have a model result to reproduce | SEAT© | Finds the prompt that produced it |
| You need to verify an AI's output | SCORE© | Has another AI rate and critique the result |