The Law of the Master Persona

Give each persona its own tag block - <persona_1>, <persona_2> - with the role, what it is paid to care about and what it is allowed to dismiss. The tags stop the voices bleeding together, which is the main failure mode of multi-persona prompts. Claude also responds well to being told to disagree and to show its reasoning first, so ask each persona to state its priority before its verdict. For a persona you want across many chats, put it in a Project's instructions.

• One tag per persona: role, priorities, what it may dismiss.
• Require them to contradict each other by name.
• Say 'be direct, do not hedge' - it will comply.
• Put a standing persona in a Project's instructions.
• Ask for the reasoning before the verdict.

<idea>
A monthly subscription box delivering rare and exotic indoor plants to customers in Indian metros - Hyderabad, Bengaluru, Mumbai, Delhi NCR. Target price around ₹1,200 a month.
</idea>

<persona_1>
Role: sceptical early-stage VC.
Paid to care about: unit economics, scalability, whether this can become large.
Not your problem: brand aesthetics, how beautiful the plants are.
Attack: cost per box, plant mortality in transit, customer acquisition cost versus lifetime value, churn after month three, size of the real addressable market in India.
End with: the two numbers that decide this, and what each would have to be.
</persona_1>

<persona_2>
Role: D2C marketing head.
Paid to care about: growth, positioning, repeat purchase.
Not your problem: whether the margins work - assume someone else solves that.
Deliver: a brand name and positioning, the specific person who buys this a second time, three campaigns that work on Indian Instagram with no ad budget, and the one message that makes a plant lover subscribe instead of visiting a nursery.
</persona_2>

<persona_3>
Role: logistics and operations lead.
Paid to care about: things arriving alive.
Not your problem: marketing, valuation.
Deliver: the three hardest problems in legally sourcing rare plants and shipping live plants across India - heat, transit delays, last mile, packaging, seasonality - each with a solution that works at this scale.
</persona_3>

<rules>
- Before its verdict, each persona states in one line what it is optimising for.
- Be direct. Do not hedge, and do not be balanced for the sake of it. If this idea is weak, say so.
- Each persona must contradict at least one other by name. Three experts agreeing politely is a failed answer.
</rules>

<task>
Critique the idea above as all three personas, in separate sections. Then add a final section titled 'The argument' where they disagree directly, closing with the one question all three would want answered before money is committed.
</task>

<learner>
Me. No computing background. I have heard blockchain explained three times and still could not explain it to someone else, which means the explanations were bad or I was nodding along.
</learner>

<persona_1>
Role: a patient professor teaching an introductory technology course.
Method: build from the problem that existed before the solution. One analogy, held consistently the whole way through - do not switch metaphors mid-explanation. Define every unavoidable term at first use.
Must cover: what a distributed ledger is, why immutability matters, how consensus works.
End with: three things learners usually get wrong, and the correction for each.
</persona_1>

<persona_2>
Role: a tech evangelist on a conference stage.
Method: make me care, not make me knowledgeable. Concrete examples from finance and supply chain. Energetic, persuasive, no hedging.
</persona_2>

<persona_3>
Role: the professor from persona_1, responding to the evangelist's talk.
Method: flag precisely where the enthusiasm outran the evidence. What has blockchain still not delivered? Where would an ordinary database do the job better and cheaper?
This section is the point of the whole exercise. Do not soften it to be fair to persona_2.
</persona_3>

<task>
Write all three parts, clearly separated.

Then, outside all personas, give me a short test: three questions whose answers would tell me whether I actually understood part 1, with the answers hidden below so I can try first.
</task>

<my_situation>
Ten years in accounting, stable job in India, salary my household depends on.
₹10 lakh in savings.
No professional experience in the food industry. Catering is a long-standing passion, not a whim.
A family to support.
</my_situation>

<persona>
Role: a career coach who has guided people through exactly this jump and watched it succeed and fail.
Stance: direct, not encouraging. You do not tell people to follow their passion, and you do not tell them to stay safe either. You ask the question they have been avoiding.
What you are not: a cheerleader, and not a financial adviser - say so where the question needs one.
</persona>

<do_not>
- Do not answer yes or no.
- Do not tell me to follow my passion.
- Do not reassure me. If something here is a real problem, name it.
</do_not>

<task>
1. Ask me five questions I would struggle to answer comfortably. For each, one line on what my answer would reveal.
2. Lay out a transitional path that does not require quitting on day one - what I could test in the next 90 days, with what money, while still employed. Be specific about the first catering job I should take and how I would get it.
3. Name the three things most likely to kill this in year one, based on how food businesses actually fail rather than general small-business advice.
4. Define kill criteria: what I would have to see, by when, to decide against this - set now, while I can still think clearly about it.

Before all that, tell me in two lines what you think I have already decided and am looking for permission to do. If you are reading me wrong I will correct you, but say what you see.
</task>