Prompt frameworks are pre-structured templates for specific kinds of work. They save cognitive effort by providing a ready structure the practitioner can fill in, and they produce consistent results across tasks of the same kind. Three frameworks cover most general professional use: CREST for general tasks, SCQA for analytical work, and DEEP for refinement. Each framework is presented below with its structure and a worked example.
CREST: Context, Role, Expectation, Structure, Tone
CREST is a compact framework for everyday professional prompts. It rearranges the five components from Section 2 into a memorable sequence that practitioners can apply quickly.
Context tells the AI what situation the task exists within. Role tells it who to be. Expectation describes what success looks like. Structure specifies how the output should be organised. Tone sets the register.
A worked example shows the framework in action. A project manager needs to produce an update memo for a client on a project that is running behind schedule. The CREST prompt might read: "Context: we are six weeks into a twelve-week project with the client, and the current assessment is that we will deliver three weeks late due to scope expansion on the first workstream. Role: you are a senior project manager known for clear communication under pressure. Expectation: the client should finish reading this memo understanding what happened, what we are doing about it, and what we need from them. Structure: four short sections, covering current status, root cause, recovery plan, and requested decisions. Tone: direct and accountable, without defensive language or excessive apology."
The resulting output reflects the framework's structure. The memo opens with current status, develops the root cause, proposes a recovery plan, and closes with specific requests. The tone is direct. The length matches what a busy client would actually read. The practitioner can then refine the draft based on the specific project circumstances, but the starting point is substantially more useful than what a simple "draft a status memo" prompt would produce.
CREST works well for memos, emails, briefs, short analyses, and most other short-form professional documents. It is the framework to reach for when the task is routine but the stakes are high enough that quality matters.
SCQA: Situation, Complication, Question, Answer
SCQA is a framework for analytical work where the practitioner needs the AI to develop structured analysis rather than produce a standard document. It comes from the consulting tradition as a structure for analytical thinking, and it translates well into a prompt structure that produces analytical output.
Situation describes the current state. Complication identifies what has changed or what the problem is. Question defines what specifically needs to be answered. Answer asks the AI to propose structured analysis.
A worked example. A financial analyst covering a pharmaceutical company needs to develop a view on whether the recent launch of a new drug justifies upgrading the stock. The SCQA prompt might read: "Situation: the company I cover reported Q3 results last week with revenues in line with consensus but with a new drug launch performing 20% above the internal forecast. The stock is currently trading at a modest premium to the peer group. Complication: the outperformance on the new drug is significant enough that my current model meaningfully understates the company's trajectory, but the market has not moved substantially on the news. Question: should I upgrade my rating, and if so, what price target adjustment does the evidence support? Answer: produce a structured analysis that considers the sustainability of the new drug's outperformance, the implications for my forecast, the comparable company valuations that would anchor a new price target, and the risks that would argue against upgrading. Conclude with a specific recommendation and the supporting logic."
The output from this prompt is substantially more useful than what a general "should I upgrade this stock" request would produce. The SCQA structure forces the analyst to be specific about the situation and the question, and it gives the AI a clear framework for producing analytical output rather than a general response.
SCQA works for consulting analyses, investment analyses, operational problem-solving, strategic evaluations, and any work where the practitioner needs structured thinking rather than standard document production.
DEEP: Diagnose, Evaluate, Enhance, Present
DEEP is a framework for refining existing work. It applies when the practitioner has a draft (whether theirs or the AI's) that needs improvement, and they want the AI to systematically work through the improvements.
Diagnose identifies what is wrong or missing. Evaluate decides what needs to change and why. Enhance rewrites or restructures to address the identified issues. Present delivers the clean revised version.
A worked example. A consultant has produced a draft executive summary for a client recommendation and wants to sharpen it. The DEEP prompt might read: "I am attaching a draft executive summary I have written for a client board recommendation. Diagnose: identify the three most important weaknesses in this draft. Evaluate: explain what each weakness does to the persuasiveness or clarity of the summary, and what specific changes would address it. Enhance: rewrite the draft to address the identified weaknesses, maintaining the specific facts and conclusions but strengthening the structure and argument. Present: deliver the revised summary in a format ready for inclusion in the broader deliverable, and include a short note below listing the changes you made so I can verify them."
The DEEP output gives the practitioner a diagnosis, a reasoning track, a revised draft, and a change log. The change log is particularly valuable because it lets the practitioner verify that the revisions are consistent with the underlying substance rather than departing from it. The framework produces useful output because it splits the refinement work into discrete stages and asks for specific deliverables at each stage.
DEEP works for refining any existing document, analysis, or presentation, and it is also useful when the practitioner wants to improve AI-generated output rather than accepting the first draft.
Using Frameworks in Combination
The three frameworks combine in useful ways. A practitioner might use SCQA to produce an initial analysis, then use DEEP to refine the analysis into its final form, then use CREST to produce the communication document that will carry the final recommendation to its audience. A practitioner working on a complex project might use SCQA at the strategy stage, CREST for each of the resulting deliverables, and DEEP whenever the draft deliverables need to be tightened before presentation.
The frameworks are templates rather than rigid structures. An experienced practitioner adapts them to the specific situation, leaving out components that do not apply and extending components that need more development. The value of the frameworks is that they provide a starting structure rather than a finished product. Practitioners who rely on them as rigid forms often produce stilted output. Practitioners who use them as scaffolding typically produce stronger work than they would produce without the scaffolding, while retaining the flexibility to adapt to the specific task.