Real Prompts You Can Steal: How Family Offices Use AI Agents on Their Wealth Data

Last updated: August 25, 2026.

Quick Answer

This article shows what AI agents can do on family office wealth data: 11 real prompts, the full answers they returned, and screenshots to prove it. Connect your structured wealth data from a platform such as Aleta to the AI assistant you already use, whether that is Claude, ChatGPT, Microsoft Copilot, or Gemini, and you can ask questions, model scenarios, and automate document work in plain language. After the demonstrations, 11 more prompts are ready to try. Every example was reproduced on demo data, and every prompt is ready to copy.

Key Takeaways

• The prompts in this article span the work a family office actually does: performance, liquidity, risk, compliance, and private markets, plus portfolio modeling and workflow automation.

• An agent connected to a wealth platform queries the office’s reconciled book of record, so every figure it retrieves is traceable back to the platform.

• In the demo environment, these prompts surfaced three breached mandates out of twelve, flagged a 35% vintage-year concentration in private equity commitments, and extracted five capital call notices matched to their instruments in Aleta.

• Modeling outputs are assistant-generated scenarios. The quality depends on the assumptions, and a human reviews anything that informs a decision.

• Every example in this article was reproduced in a demo environment with fictional data. No client data appears in any screenshot.

What Do You Need to Connect an AI Agent to Your Wealth Data?

The setup is shorter than most offices expect.

You need two things: a wealth platform that can share its structured data with AI assistants, such as Aleta, and an assistant that supports the connection, such as Claude, ChatGPT, Microsoft Copilot, or Gemini. Under the hood the connection runs on MCP (Model Context Protocol), an open standard, and the assistant works with your permissions and sees the data you can see, nothing more. That is all the technology this article requires; the plain-language walkthrough lives in MCP: What It Is and Why It Matters for Family Offices.

The prompts below are real. They come from conversations my colleagues and I have had with family offices across North America, Europe, and beyond over the past six months, and from what our own teams run day to day. They were all run on Claude Sonnet 5, and everything was reproduced in a demo environment before publishing; the entities named in the answers, the Michael Smith family, are fictional demo data.

Wealth & Performance Prompts

These are the questions that used to mean a portal session or a call to the team. Now they get answered in the time it takes to type them, with every figure drawn from reconciled data.

Prompt 1: “What changed in my total wealth this week, and why?”

Answer: "Your total wealth is up $1.54M (+0.21%) over the last 6 trading days (Thu Aug 13 – Thu Aug 20), across all four entities — Michael Smith LLC, Private Accounts, Family Smith LLC, and the Family Trust.

Why: it wasn't a steady climb. The week was choppy and driven almost entirely by one day — Wednesday the 19th added +$7.52M, more than offsetting a sharp pullback the day before (-$3.23M on Tuesday the 18th) when nearly every entity dropped at the same time. Family Smith LLC and Michael Smith LLC were the two biggest swing factors on both days.

Net effect: a volatile stretch that still closed positive, largely on the strength of Wednesday's rally."

Prompt 2: “Give me the top five contributors and detractors to this month’s performance.”

Answer: “Top 5 contributors (Aug 1–20): Apple Inc. (+$16.2M), physical gold (+$9.2M), Eli Lilly & Co (+$5.1M), Thales SA (+$2.1M), SAP AG (+$1.5M).

Top 5 detractors: Infineon Technology (-$2.3M), Koninklijke Ahold Delhaize (-$0.57M), iShares Core S&P 500 ETF (-$0.55M), Fiserv, Inc. (-$0.54M), UnitedHealth Group (-$0.35M).

One flag on Apple: the jump is unusually large relative to a normal 3-week price move, so it's worth a quick sanity-check against any recent trade activity in that position before this goes into a client-facing report — I'd rather confirm it's pure performance and not a funding/transfer effect before we lean on that number.

(Scope note: this is Michael Smith LLC, which represents the bulk of the family's public-markets exposure. Happy to extend this to the other three entities if you want the fully consolidated view.)”

Prompt 3: “Which of our accounts within the portfolios have performed best this quarter?”

Answer: “The Goldman Sachs Managed Account (under Family Smith LLC) is the top performer this quarter at +7.7%, driven largely by its Apple position. The three Schwab accounts (IRA, Brokerage, Joint Managed) are tied at +5.75% — they hold the same underlying position, which is why they move together.

The larger "core" accounts — Michael Smith LLC's main depot (+3.2%) and the Family Trust's main depot (+1.0%) — lag behind because they're far more diversified, so no single winner moves the needle as much.

Two accounts are flat: the private-assets buckets under Michael Smith LLC and Family Smith LLC show 0% because they hold private equity positions that only reprice at NAV events, not daily.

One account I'd flag separately: the Morgan Stanley Inherited IRA is showing as a negative market value (a liability, likely margin debt), so it doesn't belong on a "best performing" ranking — but it did shrink by about $337K this quarter, which is worth mentioning if the client asks about debt paydown.”

Cash & Liquidity Prompts

Liquidity is where families feel data fragmentation first: cash sits across accounts, entities, and currencies. One prompt consolidates it.

Prompt 4: “How much cash and near-cash liquidity do we have right now, across all entities and currencies?”

Answer: “You're holding $72.2M in cash and cash-equivalents across the four entities, roughly 9.8% of total AUM ($739M).

Breakdown by entity:

• Michael Smith LLC.: $57.2M (10.2% of its AUM) — by far the largest pool, includes a money-market-fund-style near-cash position

• Family Smith LLC.: $7.8M (13.4% of its AUM)

• Smith Family Trust 2018: $6.8M (5.6% of its AUM)

• Michael Smith Private Accounts: $0.35M (17.5% of its AUM — small in dollar terms but the highest relative cash weight)

On currency: virtually all of the identifiable cash sits in USD. I checked for separate EUR/GBP cash pools and didn't find meaningful ones — the EUR-denominated positions in the data are mostly private-market fund commitments (Blackstone, EQT), not cash, and the GBP "Private Assets" sub-account currently shows no cash balance. So this is effectively a single-currency liquidity position, which is worth flagging if the family has near-term EUR or GBP obligations, since they'd need to convert.”

Prompt 5: “Break down our liquidity by asset class and give me a simple cash projection based on that.”

Answer: “Consolidated asset allocation across all four entities ($739.1M total):

• Equities: $464.3M (62.8%)

• Mutual funds/ETFs: $104.2M (14.1%)

• Gold: $87.2M (11.8%)

• Cash: $72.2M (9.8%)

• Private markets: $16.3M (2.2%)

• Loans (margin liability): -$5.0M (-0.7%)

Cash projection: this is intentionally simple, and I want to be upfront about why. The only recurring cash flow I can see booked in the system is dividend/interest income (~$484K over the trailing 12 months, or roughly $40K/month). Projecting that forward flat gets you from $72.2M today to about $72.7M in 12 months — essentially flat, since there's no booked capital-call schedule, planned distributions, or discretionary spending in the data to net against it.

In other words: this projection only reflects "what happens if nothing changes." The moment you layer in actual planned capital calls (like the five we looked at earlier, which alone would draw down roughly $15.75M this fall) or any distributions, the real trajectory would look very different. I'd treat this chart as a floor, not a forecast — happy to rebuild it once real capital-call and distribution schedules are loaded into Aleta.”

Prompt 6: “Which entities are sitting on excess cash that could be put to work?”

Answer: “There's a distinction between "highest cash %" and "most actionable in dollar terms":

Most actionable: Michael Smith LLC. $57.2M in cash at 10.2% of AUM — the dollar amount here dwarfs everyone else, so even a modest reallocation (say, moving 4-5 points down toward the client's other entities' cash levels) would free up $20-25M to deploy.

Highest relative cash weight: Michael Smith Private Accounts at 17.5% — but this is only $0.35M in absolute terms, so not a meaningful deployment opportunity on its own.

Also elevated: Family Smith LLC. at 13.4% ($7.8M) — worth a look, especially since this entity also had the strongest quarter-to-date performance (+7.4%), so there could be an argument for redeploying some of that cash to keep momentum going.

Least excess: Smith Family Trust 2018 at 5.6% ($6.8M) — this looks closest to a "working" cash level already, so I wouldn't flag this one.

One caveat worth saying out loud to the client: I don't have a formally documented cash target per entity in the system (only one narrow limit exists — a 5% cap on cash within the Schwab sleeve specifically, and that one's currently fine at 2%). So "excess" here is relative to each other and to typical family-office norms, not a breach of a stated mandate. If the family has house-view cash targets per entity, I'd recommend we load those as formal limits so this becomes a compliance check rather than a judgment call.”

Risk & Concentration Prompts

Concentration builds through price appreciation long before anyone decides on it. Spotting it usually takes a dedicated analysis. Here it takes a sentence.

Prompt 7: “How concentrated are we in single-name positions right now?”

Answer: “Your top 10 positions make up 55.6% of Michael Smith LLC — over half the portfolio sits in just ten names. Three positions clear the common 5% single-name threshold: physical gold at 14.6%, Eli Lilly at 11.2%, and Thales SA at 6.3%.

The gold position is the standout — 14.6% in a single commodity holding is unusually high for a diversified family office book, and worth confirming whether it's a deliberate strategic allocation or something that's grown into an outsized weight through price appreciation without rebalancing. Beyond the top three, there's also a cluster worth noting: healthcare names (Eli Lilly, plus Polar Capital Biotech and Wellington Healthcare further down) and European industrials/tech (Thales, SAP, Infineon) each show up more than once in the top 10, which adds a layer of sector concentration on top of the single-name picture.

(Scope note: this is Michael Smith LLC only. Happy to extend to the fully consolidated view across all four entities if you want to check for the same name showing up in more than one account.)"

Compliance Prompts

Mandate checks are the kind of work that slips down the list until quarter-end. An agent turns them into a question you can ask any morning.

Prompt 8: “Are we within our strategic asset allocation limits, or are any mandates being breached?”

Answer: “Yes — 3 of your 12 active mandates are currently breached, all on the equities side:

1. Equities target: 62.1% actual vs. 40% max — the biggest overall issue, you're running 22 points over the strategic equity ceiling.

2. USD equities exposure: 62.1% vs. 50% max — a related breach, largely the same underlying overweight.

3. Schwab American Eq: 46.8% vs. an 80–100% mandated band — this one's the opposite problem: within the Schwab sleeve specifically, you're required to hold mostly American equities, and you're currently well under that floor.

The other 9 mandates — including bonds allocation, European equities SAA, Schwab cash, and the various country and consolidated-group limits — are all within range.

Reading these together: the portfolio has drifted meaningfully overweight equities at the whole-of-family level, while the Schwab account specifically has drifted away from its US-equity mandate even as the broader book gets more equity-heavy elsewhere. That's a rebalancing conversation worth having — likely trimming equity exposure overall while adding US names specifically within Schwab.”

Private Markets Prompts

Private markets data is the hardest to query by hand: commitments, vintages, and calls sit across documents and entities. The structured foundation underneath is what makes a prompt like this one possible.

Prompt 9: “How diversified are we across vintage years in our private equity commitments?”

Answer: “You have $195M committed across 9 funds, but it's not evenly spread — there are two issues worth flagging:

Concentration in 2021: $68.5M, or 35% of total commitments, all went into one vintage year (Blackstone Capital Partners VIII, KKR Americas XII, and Bain Capital Fund XII all closed in 2021). That means over a third of your private markets book is exposed to the same entry-point market conditions and will likely hit exit windows around the same time.

A two-year gap in 2023–2024: no new commitments at all. Vintage diversification exists specifically to smooth out the "buy at one point in the cycle" risk — skipping two consecutive years means you missed capturing whatever pricing/opportunity existed in that window, and it widens the gap between your 2022 and 2025 fund generations.

The rest is reasonably spread: $35M in 2019 (CVC), $42.5M in 2020 (Advent, Silver Lake), $43M in 2022 (EQT, Carlyle), and a small $6M 2025 vintage (Aleta Capital Fund VII) that looks like the start of a new commitment cycle.

Bottom line: if this were a real client conversation, I'd suggest the next 1-2 new fund commitments intentionally target 2026 vintage to keep building out that newer cohort, rather than adding to the already-heavy 2021 exposure.”

Portfolio Modeling Prompts

This is the part of every conversation where people lean in. Give an assistant access to real positions and it projects them forward, charts and scenario tables included. Two things I always add when I show it: the outputs are scenarios that stand or fall with their assumptions, which the assistant below states and flags on its own, and they complement the model-based private markets forecasting inside the platform itself. Treat them as decision support, reviewed by a person, never as advice.

Prompt 10: “Run a Monte Carlo simulation on my portfolio. Account for a 10-year time period with a bull case, a bear case, and a base case, and show me some visualizations.”

Answer: “Ran 20,000 simulated 10-year paths starting from your current $739.1M, using standard geometric Brownian motion:

Workflow Automation Prompts

This is where the horsepower is. The agent stops answering questions and starts completing work: reading documents, extracting what matters, and preparing entries for the platform, with your approval before anything posts.

Prompt 11: “Take these 5 capital call notices, extract the necessary information, and tell me what you'd post to Aleta before actually doing it.”

Answer: “Here's what I extracted from the 5 capital call notices, matched against your actual private-equity fund holdings in Aleta: