Advanced Prompting for Complex Planning Cases

The simple Role-Context-Task-Format prompt handles most of what you need. But the genuinely hard cases — a business owner contemplating a sale, a blended family with competing estate goals, a near-retiree weighing a pension lump sum against an annuity — reward more deliberate technique. This lesson covers the advanced moves: structured context, step-by-step reasoning, multi-perspective analysis, and chaining.

What You'll Learn

• Feeding complex situations to AI without overwhelming it (or yourself)
• Prompting for explicit, step-by-step reasoning you can audit
• Getting multiple expert perspectives on one situation
• Chaining prompts so a big problem becomes a sequence of small ones

Structured Context for Messy Situations

Complex cases have a lot of moving parts. Dumping them in as a paragraph gets you a mushy answer. Structure the input.

You are an experienced financial planner helping me think through a complex case (all anonymized — no names, no account numbers). I'll give you the situation in a structured format. After I do, ask me up to 5 clarifying questions before you analyze anything.

People: Owner, 58, and spouse, 56. Two adult kids from owner's first marriage; one minor child together. Business: Owner runs a services company; considering a sale in 2–4 years; rough enterprise value range [generalized]. Assets: Retirement accounts [generalized total], taxable brokerage [generalized], home equity [generalized], the business. Income & spending: [generalized]. Goals: Retire shortly after the sale; maintain current lifestyle; fund the minor child's education; "be fair" to all three kids in the estate plan; minimize taxes on the sale. Concerns: Concentration in the business; estate tension between the older kids and the spouse/minor child; what to do with sale proceeds. Constraints: I do the actual recommendations and any binding numbers; you help me think and frame.

Asking it to ask you clarifying questions first is a powerful move — it surfaces the gaps before it commits to an analysis, and it mirrors how a thoughtful colleague would approach it.

Make the Reasoning Explicit

For anything with logic to it — a decision, a comparison, a trade-off — ask the AI to show its work. Reasoning you can read is reasoning you can check.

Walk through this step by step. First, restate the key facts and what we're optimizing for. Second, lay out the main options for the sale-proceeds question. Third, for each option, the pros, cons, and what would have to be true for it to be the right call. Fourth, the questions or data I'd need to settle it. Don't jump to a conclusion — I want the structured thinking, not a verdict. If any step requires a calculation, describe what should be calculated and flag that it needs to be done in planning software, not estimated by you.

This does two things: it gives you a usable framework for the client conversation, and it lets you spot where the AI's logic is off (it sometimes is) before that ever matters.

Multiple Perspectives on One Problem

Hard cases benefit from being looked at through several lenses. Ask for them explicitly.

Analyze this situation from four perspectives, separately and clearly labeled:

1. The retirement-income planner — does the post-sale picture support the lifestyle, and what are the risks?
2. The tax-aware lens — what are the big tax considerations around the sale and the proceeds, at a conceptual level, and where would a CPA need to be involved?
3. The estate-planning lens — how do you balance "fair to all three kids" with providing for the spouse and minor child, and what tools/structures are commonly used (conceptually)?
4. The behavioral / family-dynamics lens — where is this likely to get emotional, and how should I, as the advisor, handle those conversations? For each, end with the top two questions I should be asking. Keep it conceptual — flag clearly where other professionals (CPA, estate attorney) must be brought in.

You're not getting advice to act on; you're getting a structured map of the terrain so you, the CPA, and the estate attorney can do your jobs well.

Chaining: One Big Problem, Many Small Prompts

Don't try to solve a complex case in a single mega-prompt. Break it into a chain, where each step builds on the last:

1. Frame: "Summarize this situation and what we're optimizing for; list the major decision points." → you review and correct.
2. Explore one decision: "Take decision point #1 — the business-sale structure question, conceptually. Lay out the common approaches and trade-offs." → you review.
3. Client-facing translation: "Now explain decision point #1 to the client in plain English — the choice they're facing and the trade-offs — without recommending one." → you edit.
4. Conversation prep: "Draft talking points and likely client questions for the meeting where we discuss this." → you finalize.
5. Documentation: "Summarize where we landed and the open items for my notes." → you check and file.

Each step is small, reviewable, and correctable. If step 2 goes off, you catch it before it contaminates steps 3–5. This is how you use AI on the cases that matter without letting it run away with them.

The Advanced Rules Are the Same Rules

More sophisticated technique doesn't loosen the constraints — if anything it raises the stakes:

• Still anonymized, still firm-approved tool. Complex cases have more identifying detail, not less — be extra careful.
• Still no binding math. The harder the case, the more tempting it is to let the AI "just estimate" — don't. Planning software and the CPA.
• Still no recommendations from the AI. It maps options; you (with the CPA and attorney) decide.
• Still your review on everything. Especially here — the AI's reasoning is more likely to have a subtle flaw on a complex case, and the cost of missing it is higher.
• Know when to bring in the humans. A sophisticated AI analysis is not a substitute for a CPA or an estate attorney. Use it to prepare for those conversations, not replace them.

Key Takeaways

• For complex cases, feed the situation in a structured format and ask the AI to ask you clarifying questions before analyzing — it surfaces the gaps a thoughtful colleague would catch.
• Ask the AI to show its reasoning step by step rather than jump to a verdict; auditable reasoning gives you a usable framework and lets you catch logic errors.
• Request multiple expert perspectives (retirement-income, tax-aware, estate, behavioral) on one situation, with clear flags for where a CPA or attorney must be involved.
• Chain a big problem into small, reviewable steps — frame, explore one decision, translate for the client, prep the conversation, document — and remember the advanced rules are the same rules: anonymize, no binding math, no AI recommendations, review everything, bring in the human experts.