How to Write an IPS with AI Assistance

Investment Policy Statements take advisors hours to draft, longer for complex situations. AI turns most of that drafting into editing, and makes the documents more consistent across clients. Here's how.

What AI handles in IPS drafting

The structured, repeatable parts:

• Client objective summary based on planning data
• Risk tolerance framing tied to risk profiling output
• Asset allocation targets and rebalancing rules
• Liquidity needs and time horizon analysis
• Tax considerations and account structure
• Standard disclosure and legal language

What the advisor handles:

• The strategic call on targets and policy
• Client-specific judgment (concentrated holdings, restricted positions, special circumstances)
• Final review and sign-off
• Client conversation explaining the IPS

The inputs the AI needs

Before any drafting, collect:

• Client objectives (from planning software or discovery meeting notes)
• Risk tolerance output (from Nitrogen, Tolerisk, or similar)
• Income, expense, and cash flow profile
• Account structure (taxable, IRA, Roth, Trust, etc.)
• Time horizon and major financial milestones
• Concentrated holdings or restrictions
• Tax situation (marginal rate, expected liquidity needs)
• Firm IPS template (as the baseline)

The data lives in your planning software (eMoney, RightCapital, MoneyGuidePro), CRM, and risk profiling tool. Pull it once, prompt the AI, generate the IPS.

The prompt scaffold

Draft an Investment Policy Statement for a new client based on the following inputs.

CLIENT PROFILE
- Name: [client name]
- Age: [X], retirement age target: [Y]
- Current AUM: $[X]
- Marginal tax rate: [%]

OBJECTIVES
[from planning meeting]

RISK PROFILE
- Risk Number (Nitrogen): [X]
- Time horizon: [years]
- Drawdown tolerance: [verbalized comfort with X% peak-to-trough]

ACCOUNT STRUCTURE
- Taxable: $[X]
- Tax-deferred (IRA): $[X]
- Tax-free (Roth): $[X]
- Trust/other: $[X]

TARGET ALLOCATION (advisor-set)
- Equities: [X]% (US: Y%, International: Z%)
- Fixed Income: [X]% (Core: Y%, Munis: Z%)
- Alternatives: [X]%
- Cash: [X]%

LIQUIDITY NEEDS
[any near-term cash needs, distributions planned]

SPECIAL CONSIDERATIONS
[concentrated positions, restrictions, ESG preferences, etc.]

OUTPUT FORMAT
Use our firm's standard IPS template structure:
1. Purpose and authority
2. Investment objectives and time horizon
3. Risk tolerance and asset allocation
4. Rebalancing policy (annual + drift threshold of [X]%)
5. Tax considerations and account placement
6. Roles and responsibilities
7. Monitoring and review (frequency: [quarterly/annual])
8. Disclosures

Voice: clear, plain English where possible, technical where required. Avoid jargon when a plain word works.

Length: 6-10 pages standard, longer if special circumstances warrant.

Include firm boilerplate disclosures at the end (paste from firm template).

What you get back

A complete IPS draft to read and edit. The editing time depends on the complexity of the household.

The draft will be:

• Structurally complete (all sections present)
• Voice consistent with the firm template
• Tied directly to the client's specific inputs

What you'll still need to edit:

• Client-specific framing and emphasis
• Any unusual situations the prompt didn't fully capture
• Final disclosures and legal language to firm standard
• Specific language compliance has flagged in the past

Compliance considerations

IPS is a books-and-records document. AI-drafted IPS is the same as advisor-drafted IPS from a regulatory perspective — same review, same retention, same supervisory standards.

The clean model:

• AI drafts
• Advisor reviews and edits
• Compliance reviews per firm policy (typically once per new IPS or major change)
• Client signs (or acknowledges)
• Filed in firm records
• Reviewed annually or per firm IPS-review policy

AI changes the draft step. It does not change the review, sign-off, or retention process.

The mistakes to avoid

Mistake 1: Letting AI define targets. AI should reflect targets the advisor has decided. It should not invent or recommend targets. If the prompt doesn't include explicit targets, the AI will produce something — but it's not from a fiduciary decision.

Mistake 2: Skipping the planning input. Without real planning data, the IPS reads generic. The quality of input determines the quality of output.

Mistake 3: Trusting boilerplate. AI tends to use generic disclosure language. Always replace with your firm's exact approved boilerplate.

Mistake 4: Single-pass drafting. The first draft is a draft. Plan real editing time for client-specific voice and anything unusual in the situation.

How to size it

Time the next few IPS drafts written by hand, then the same work with the AI draft. Multiply the difference by the number of new clients you onboard in a year and by what an advisor hour is worth to the firm. The other gain is consistency across the firm. Every client gets a thorough document, not a thin one because the advisor was time-pressed.

When to skip AI on IPS

Two situations where manual still beats AI:

1. Highly complex clients with unusual structures — Multi-generational trusts, business succession, concentrated executive comp. The structure is too specific for prompt-based drafting; build manually.
2. First IPS at a new firm — Spend time getting the template right manually. Then use AI to scale.

Bottom line

IPS drafting is repetitive structured work with high consistency requirements. That's exactly the work AI handles well when the inputs are clean. Built right, the workflow turns drafting into editing, improves consistency, and frees advisor time for the actual client conversation about strategy.

The IPS isn't replaced — its drafting step is. The advisor still decides, edits, signs.