Ten prompts, one scaffold. AI Central's consultant set casts ChatGPT as a strategy consultant, productivity consultant, career coach, marketing advisor, finance planner, fitness coach, content strategist, negotiation coach, decision advisor and problem-solver. Every one is built the same way: assign the role, load real detail into bracketed slots, name the analysis you want, then dictate the shape of the answer. That last instruction is the one most people drop, and it is the one that decides whether you get advice or a deliverable.
Reviewed August 2026.
What the set is doing that a prompt list usually is not
Most prompt collections are a pile of one-liners with nothing holding them together. AI Central's set is one template applied ten times. Read the prompts back to back and the repetition stops looking like laziness and starts looking like a specification.
Each one opens with the same five words, I want you to act, and then names a consultant rather than a task. That is a deliberate choice. Asking ChatGPT to help with your marketing gets you a list of tactics. Telling it to act as your marketing consultant, then handing it a budget, a set of channels and an audience, gets you an audit, because an audit is what a consultant produces when handed a client situation.
The four moves inside every prompt
Strip the topic away and the same four instructions are left in all ten.
- Role. One sentence that assigns a job title, not a task.
- Context. Bracketed slots for the facts only you have, your role, your hours, your budget, your income, your constraints.
- Analysis. A named operation to run on that context, audit it, compare it, break down the root causes, run a skill gap analysis.
- Format. A closing sentence that dictates how the answer arrives, as a table, a script, a weekly breakdown, a calendar.
The order matters more than it looks. Role before context tells the model what to pay attention to when it reads your numbers. Analysis before format stops it optimising for a tidy table before it has done any thinking.
The brackets are the actual work
These prompts are useless empty. Every one of the ten is mostly placeholders, and the quality of the answer tracks almost perfectly with how specific you were in filling them.
Look at how much the set asks for. The fitness prompt wants age, weight, height and fitness level, plus constraints such as thirty minutes a day, a vegetarian diet, or no gym access. The finance prompt wants monthly income, fixed expenses, current debts and assets, and a dated savings target. The productivity prompt wants your role, your hours per week, your five recurring tasks and your bottlenecks.
None of that is padding. A consultant with no client data gives generic advice, and so does a language model. Every bracket you leave vague converts directly into a paragraph of filler in the output.
The strategy prompt is explicit about where it wants to land, in AI Central's words:
End with a prioritized action plan tailored to my situation.
Tailored to my situation is the load-bearing phrase there. It only means anything if the situation was actually described.
The format sentence is a contract, not decoration
Here is what separates this set from the average prompt roundup. All ten prompts close with an output instruction, and counted across the ten as written, no two of those instructions are the same.
The strategy prompt asks for clear headings and bullet points. Productivity asks for a clear, actionable list. Career asks for a section per component. Marketing asks for bold headings on each strategy. Finance asks for clear categories and a table for the budget. Fitness asks for output structured by week with daily instructions. Content asks for ideas in a table and the schedule in a calendar. Negotiation asks for a script followed by key strategies. Decision-making asks for the matrix as a table with the recommendation stated clearly. Problem-solving asks for the analysis first and the solutions second.
Ten prompts, ten different shapes, each matched to what you would actually do with the answer. A budget wants a table because you are going to compare rows. A negotiation wants a script because you are going to say the words out loud.
Drop that final sentence and ChatGPT defaults to essay mode, and you spend the next ten minutes reformatting by hand.
It names frameworks instead of hoping for them
The set does something quietly smart with methodology. Rather than asking for good advice, several prompts name the method they want applied.
The productivity prompt names GTD, time-blocking and the Eisenhower Matrix as examples of the systems it wants recommended. The problem-solving prompt names the 5 Whys and SWOT analysis as the tools for root cause work. The decision prompt does not simply ask for a recommendation, it specifies the axes of the matrix, ROI, risk level, opportunity cost, and alignment with long-term goals. The strategy prompt sets its own comparison columns too, pros, cons, potential ROI, required resources, and execution difficulty.
Naming the framework is cheap insurance. It narrows the model onto a known structure with known outputs, which beats hoping it picks a sensible one on its own.
Where this stops working
Two honest limits, neither of which the prompts announce.
First, the model only knows what you typed. It cannot see your calendar, your bank statement or your analytics. Every plan it returns is built on your self-report, and self-reports run optimistic. If the productivity answer feels wrong, check the hours you claimed before you blame the output.
Second, the finance and fitness prompts ask for plans in areas with real personal consequences, and neither instructs the model to flag its own uncertainty or say when a professional should be involved. Treat those two as a first draft to take to someone qualified, not as a ruling.
There is also a volume trap. Several prompts request a quantity, three to five pathways, three marketing strategies, ten content ideas. Ask for a number and you will get that number, so the last entries are often padding. Read from the top down and stop when it stops being useful.
How to run the set in one sitting
Fill the brackets in a plain text file before you open ChatGPT. It takes ten minutes, and it is the only part genuinely your job.
Then chain the prompts rather than running them in isolation. The strategy prompt produces several pathways compared on pros, cons and required resources. Feed the top two into the decision prompt as option A and option B, and the matrix runs on real alternatives instead of hypotheticals. Point the problem-solving prompt at whatever that matrix flags as the biggest risk.
Finish where the set finishes. The problem-solving prompt closes with an instruction worth stealing for every other prompt you ever write:
End with the most practical recommendation I can test this week.
That one line converts a wall of analysis into a single action, and it is the difference between a document you save and a thing you do.
AI Central's closing note on the set is blunt about how the work splits:
ChatGPT is a super-effective Tool.
The prompts are the engine. Nothing moves until you fill them in and turn it over.
Do I have to fill in every bracket?
Yes, and filling one vaguely counts as not filling it. The brackets are the only thing separating a consultant answer from a blog post. If you truly do not have a number, use your best honest estimate and label it as one.
Should I run all ten prompts in one chat or start fresh each time?
Start fresh for unrelated domains, so the fitness context does not bleed into the marketing answer. Keep one chat open when you are deliberately chaining, for example carrying pathways out of the strategy prompt into the decision prompt as the two options.
Will ChatGPT actually obey the formatting instruction?
Usually, and when it drifts the fix is one message. Ask it to reformat the previous answer into the shape you specified. Keep the format sentence at the very end of the prompt, which is exactly where all ten of these put it.
Are these safe to use for money and health decisions?
Use them to get organised, not to get a verdict. The finance and fitness prompts produce a structured plan fast, which is genuinely useful, but neither the prompt nor the model knows your full circumstances. Take the output to a qualified professional before acting on anything with real downside.
What do I do when the answer comes back generic?
Look at the input, not the model. Generic output almost always traces back to an empty bracket, a goal with no number in it, or a constraint you left out. Add the specifics and re-run the same prompt.