AI Prompts for Evaluating Art, Collectibles and Passion Assets

Buying art or collectibles as an investment typically requires 55–65% appreciation over ten years just to break even, given buyer's premiums of 26–28% at major auction houses like Sotheby's, seller's commissions around 15%, and additional costs, according to a new guide on using AI prompts for evaluating passion assets. The guide stresses that AI cannot authenticate or attribute works and should only be used to interrogate provenance, model costs, and prepare questions for specialists.

AI Prompts for Evaluating Art, Collectibles and Passion Assets Short answer: Buy it because you want to own it. If the investment case is what persuades you, the fees will eat the case — a buyer’s premium of 26–28% at major houses, seller’s commission around 15% , plus insurance and conservation, means a ten-year hold typically needs 55–65% appreciation just to break even . And one hard rule before anything else: AI cannot authenticate or attribute a work. It will produce confident, invented answers. Use it to interrogate provenance, model true cost, and prepare questions for real specialists. TL;DR — Key Takeaways The round trip is brutal. Published analysis puts total transaction costs at 15–40% of hammer price. Model it before you bid, not after. Forged provenance is as dangerous as a forged object. Fabricated photographs, receipts and ownership histories have fooled specialists for years. The 1933–45 gap is the risk that grows. Restitution mechanisms are getting easier to use, not harder. Even a short unexplained gap can create liability. Test the exit before you buy. Illiquidity is discovered at sale. The useful question isn’t “what will it be worth” but “what would make this unsellable”. AI’s role is questions, not answers. Provenance interrogation, cost modelling, insurance gaps — never attribution. ✔ Best for Newly liquid buyers entering the market, existing collectors reviewing insurance and exposure, and anyone being offered a piece with an unusually good story attached. ✕ Skip if You need a valuation for probate, insurance or sale — that requires a qualified appraiser — or you’re evaluating art funds as a financial product, which needs regulated advice. ⚠ The one hard rule: AI cannot authenticate. A language model cannot examine a physical object, cannot perform technical analysis, cannot see brushwork, and has no reliable access to catalogues raisonnés or foundation records. Asked whether a work is genuine, it will produce a fluent, confident, invented answer — and in this field a confident wrong answer is expensive and occasionally unlawful to act upon. Authentication requires recognised specialists, technical examination, and where one exists, the artist’s foundation or catalogue raisonné authority. Nothing in this guide substitutes for that. Every prompt below is designed to help you ask better questions of real experts — not to answer those questions itself. On this page What does it actually cost to own? Far more than the hammer price, and the gap is where most investment cases die. The buyer’s premium alone puts you materially underwater the moment the hammer falls. | Cost | Typical level | Notes | |---|---|---| Buyer’s premium | ~26% first tranche; 28% at Sotheby’s after Feb 2026 | Falls on higher tranches — around 14.5% above roughly $6M. Sales tax or VAT usually charged on the premium too. | Seller’s commission | ~15% typical 0–25% negotiable | Often waived or reduced for genuinely valuable consignments. Assume you’ll pay it. | Sale-side charges | Photography, cataloguing, LDL insurance, transport | Individually small, collectively material on mid-value works. | Insurance | Annual, as a percentage of value | Recurring for the whole holding period. Compounds quietly. | Conservation & environment | Periodic | Climate control, framing, occasional treatment. Neglect shows up as a condition discount at sale. | Published analysis puts the total round trip at 15–40% of hammer price . That range is wide because premiums are tiered and commissions negotiable — which is precisely why you should model your numbers rather than accept a rule of thumb. Why the investment case usually fails Because the friction is front-loaded and the appreciation isn’t. Work the arithmetic on a mid-value purchase held ten years and the required break-even appreciation lands somewhere around 55–65%. Round-trip friction 15–40%of hammer price, per published analysis Break-even need ~57%appreciation on a 10-year hold worked example below Annualised ~4.6%every year, just to stand still None of that means don’t buy. It means buy for possession and treat any appreciation as a bonus . The collectors who do well financially are usually the ones who bought what they loved, held for decades, and had the good fortune of taste that aged well. The ones who do badly bought an asset class. The three risks that destroy value Risk 1 Attribution Is it by whom the seller says? Downgrade from an artist’s hand to “workshop of” or “after” can remove most of the value overnight. Risk 2 Title Do they have the right to sell it? Gaps in ownership history — particularly 1933–45 — can surface as claims decades later. Risk 3 Condition Over-restoration, relining, retouching. Often invisible to the untrained eye and heavily discounted by the market. The risk most people underrate: forged provenance. Buyers concentrate on authenticating the object and take the paperwork at face value. But fabricated photographs, invented collector histories and forged receipts have deceived specialists for extended periods — the Beltracchi case being the well-documented modern example, and Van Meegeren’s invented Italian collection the classic historical one. A convenient story explaining why a work never appeared in the scholarly literature is itself a warning sign, not a reassurance. Get the Passion Asset Acquisition Checklist The full pre-purchase checklist, the total-cost-of-ownership model, the provenance question set, red-flag patterns, and the insurance adequacy audit. Free. Prompt 1: The total cost of ownership model Run this before you bid, not after you own it. The output isn’t a valuation — it’s the appreciation you’d need simply to get your money back. THE PURCHASE Category: painting / watch / classic car / wine / etc. Hammer or asking price: AMOUNT + CURRENCY Buying via: AUCTION HOUSE / DEALER / PRIVATE SALE Buyer's premium quoted: % — check the actual tier Sales tax / VAT position: APPLICABLE RATE, AND WHETHER ON THE PREMIUM Import duty if cross-border: IF APPLICABLE HOLDING Expected hold: YEARS Insurance quoted: ANNUAL AMOUNT OR % OF VALUE Storage / environment: ANNUAL COST, IF ANY Expected conservation: TREATMENT COST + FREQUENCY Transport / installation: ONE-OFF EXIT ASSUMPTIONS Likely sale route: AUCTION / DEALER / PRIVATE Seller's commission expected: % Other sale costs: PHOTOGRAPHY, CATALOGUING, LDL, TRANSPORT YOUR TASK — formulas only, do not calculate 1. THE COST STACK — lay out every cost as a formula I can put in a spreadsheet. Separate one-off purchase costs, recurring holding costs, and exit costs. 2. TOTAL CASH OUT at purchase hammer + premium + tax . 3. BREAK-EVEN FORMULA — the sale hammer required to recover everything, given the seller's commission is applied to that hammer. Show the algebra, since this is circular and people get it wrong. 4. REQUIRED APPRECIATION — as a percentage of my purchase hammer, and annualised over my hold period. 5. SENSITIVITY — give me the grid to build: premium rates down the side, seller's commission across the top, with required appreciation in each cell. 6. WHAT I'VE FORGOTTEN — costs I haven't listed that apply to this category specifically. Ask me about anything material rather than assuming. 7. THE HONEST FRAMING — state plainly what annual return this asset must produce to match a simple alternative, and note that I should treat that comparison as illustrative rather than as investment advice. Do not calculate. Do not estimate what the piece might be worth in future — you have no basis for that. Prompt 2: Provenance interrogation The goal is a question list for the seller, not a verdict. Provenance is a chain of custody, and what matters is whether every link can be independently verified — and what sits in the gaps. THE WORK: CATEGORY, ATTRIBUTED MAKER, APPROXIMATE DATE PROVENANCE AS PRESENTED TO ME: PASTE EXACTLY WHAT THE SELLER PROVIDED — names, dates, collections, sales, exhibitions, literature references DOCUMENTATION OFFERED: WHAT PHYSICAL/DIGITAL EVIDENCE EXISTS — receipts, photographs, letters, certificates SELLER TYPE: MAJOR HOUSE / ESTABLISHED DEALER / PRIVATE / ONLINE / UNKNOWN Build my interrogation list. Do NOT assess authenticity — you cannot, and I'm not asking you to. A. THE CHAIN — lay out the ownership history as a timeline. Mark every period that is UNACCOUNTED FOR, with the exact years missing. B. THE 1933–45 QUESTION — flag any gap, however short, falling in or overlapping this period, and explain why it requires specific enquiry. Note that restitution mechanisms in several jurisdictions have become easier to use over time, so this exposure does not fade. C. VERIFIABILITY — for each link in the chain, what independent source could confirm it named institution, published catalogue, auction record, exhibition history ? Mark each: INDEPENDENTLY VERIFIABLE / VERIFIABLE ONLY VIA THE SELLER / UNVERIFIABLE. The second category is the one to worry about. D. THE QUESTIONS — write the exact questions to put to the seller, in order. Polite, specific, answerable. Include the questions a specialist would ask that I wouldn't think of. E. WHAT SHOULD EXIST BUT DOESN'T — for a work of this category and claimed history, what documentation would normally exist? Absence of the expected is more telling than presence of the unusual. F. WHO TO ASK — which specialists, foundations, catalogue raisonné authorities or databases should be consulted for this maker and period. Name categories of authority; tell me to verify current contacts myself. Rules: - Do not state or imply any view on authenticity. - Do not invent provenance details, sale records or literature references. If you don't know, say so. - Flag anything that needs a specialist lawyer rather than a specialist in the field. Prompt 3: Pattern-level red flags This assesses the circumstances of a sale, not the object. Known warning patterns are recognisable without any expertise in the work itself — and they’re where a non-specialist buyer has genuine leverage. HOW THIS CAME TO ME: Describe the circumstances honestly: who approached whom, how it was framed, what story accompanies it, what timeframe I've been given, how the price compares to what I understand of the market WHAT I'VE BEEN TOLD: The narrative — the discovery story, why it's available, why now, why me WHAT'S BEEN OFFERED / REFUSED: Technical examination? Independent inspection? Time to consult? Conditions attached? Assess the CIRCUMSTANCES against known warning patterns. Do not assess the object — you cannot see it. 1. PRESSURE — any urgency, exclusivity or scarcity framing? Legitimate sales survive delay; explain why that matters. 2. THE STORY — does the narrative conveniently explain absence from the scholarly record? Historically that pattern has featured in significant deceptions, so it warrants scrutiny rather than reassurance. 3. VERIFICATION FRICTION — has independent examination been permitted, discouraged, or made impractical? Note that reluctance can have innocent explanations and should prompt a question, not a conclusion. 4. PRICE POSITION — if notably below comparable works, what would explain that legitimately, and what wouldn't? 5. THE COUNTERPARTY — what should I establish about who I'm dealing with, and what recourse exists if attribution is later challenged? 6. MY QUESTIONS — the specific things to ask, and what a satisfactory answer sounds like versus an evasive one. 7. WALK-AWAY TRIGGERS — what should end my interest outright, decided now while I'm calm. Rules: - These are patterns, not proof. State clearly that identifying a pattern does not mean anything is wrong. - Do not accuse or imply wrongdoing by any party. - Recommend independent professional examination as the resolution to every open question. Prompt 4: The insurance gap audit Most collections are underinsured in a way the owner discovers only after a loss. The common failure is a general contents policy with a per-item limit far below the value of individual works. MY CURRENT POSITION: Policy type: GENERAL CONTENTS / SPECIALIST FINE ART Basis of cover: AGREED VALUE / MARKET VALUE / UNKNOWN Per-item limit: AMOUNT, IF ANY Total collection cover: AMOUNT Last professional valuation: WHEN, BY WHOM Where items are kept: HOME / STORAGE / ON LOAN / MIXED Highest-value single item: APPROXIMATE VALUE Audit for gaps. Give me questions for my broker. A. THE BASIS PROBLEM — explain agreed value versus market value in plain terms, and which exposes me to argument at the point of claim. B. LIKELY EXCLUSIONS — what does cover of this type commonly exclude or limit? Consider: transit, items on loan, damage during restoration, gradual deterioration, inherent vice, and depreciation after repair. C. THE DEPRECIATION QUESTION — if a work is damaged and restored, its market value typically falls even after competent treatment. Does my cover address that loss? This is the gap owners most often discover too late. D. VALUATION CURRENCY — how recent must valuations be for cover to hold, and what happens if values have moved since? E. THE QUESTIONS FOR MY BROKER — written out, specific, in priority order. F. WHAT NEEDS A SPECIALIST — where a fine art insurer or specialist broker is warranted rather than a generalist. Do not recommend specific insurers or products. Do not interpret my policy wording — that requires reading the actual document. A real cost model, run in full A mid-value painting bought at auction, held ten years. All inputs are the buyer’s own; the model calculates nothing — it supplies the formulas. Hammer: $180,000 · Buyer's premium: 26% Hold: 10 years · Insurance: 0.5% of value annually Environment/care: $400/yr · Conservation: one treatment, $3,500 · Exit: auction, seller's commission 12%, sale costs $2,500 | Line | Formula | Amount | |---|---|---| | Hammer | H | $180,000 | | Buyer’s premium | H × 0.26 | $46,800 | Cash out at purchase | H × 1.26 | $226,800 | | Insurance, 10 yrs | Value × 0.005 × 10 | $11,340 | | Environment/care | 400 × 10 | $4,000 | | Conservation | one treatment | $3,500 | Total holding cost | $18,840 | | | Sale costs fixed | $2,500 | | Break-even sale hammer | 226,800 + 18,840 + 2,500 ÷ 1 − 0.12 | $281,977 | The finding: to recover the money, the work must sell at a hammer of roughly $282,000 against a purchase hammer of $180,000 . That is 57% appreciation over ten years — about 4.6% annually — simply to break even. Sensitivity, as the model laid out: | Buyer’s premium | Seller’s commission 10% | Seller’s commission 15% | |---|---|---| | 26% | 53% appreciation needed | 62% needed | | 28% | 55% needed | 65% needed | What the model flagged as forgotten: sales tax or VAT on the premium jurisdiction-dependent, potentially several thousand , import duty if buying cross-border, and the cost of a professional valuation to keep insurance cover valid — which recurs. Note what the model did and didn’t do. It never estimated what the painting might be worth in ten years — it has no basis for that, and said so. It produced the threshold, which is the decision-relevant number. If you look at 57% over a decade and still want it, buy it with clear eyes. That’s a good outcome for this prompt. Level-up: what would make this unsellable? This is the part no acquisition checklist covers. Buyers ask what will it be worth — a question nobody can answer. The useful question is the inverse: what would make this difficult or impossible to sell later? Illiquidity is discovered at exit, when it’s too late to price it in. Act as the specialist who will be asked to sell this piece in fifteen years. Your job is to identify, now, every reason you might have to decline the consignment, discount it heavily, or struggle to place it. THE PIECE: CATEGORY, ATTRIBUTED MAKER, PERIOD, MEDIUM/ MATERIALS, SIZE OR FORMAT PROVENANCE AS PRESENTED: SUMMARY, INCLUDING KNOWN GAPS CONDITION AS DISCLOSED: INCLUDING ANY RESTORATION WHAT I'M PAYING: AND HOW THAT COMPARES TO RECENT COMPARABLES, AS FAR AS I KNOW DOCUMENTATION I'LL HOLD: WHAT I'LL ACTUALLY RECEIVE Six sections: 1. THE TITLE OBJECTIONS — what in the ownership history would a future auction house's compliance team query? Be specific about which gaps require explanation and what evidence would close them. Note that standards have tightened over time and may tighten further. 2. THE ATTRIBUTION FRAGILITY — what would have to change in scholarship for this attribution to be revised downward? How dependent is the current attribution on a single opinion, a single publication, or a single authority that may not exist in fifteen years? 3. THE CONDITION TRAJECTORY — how will this material age? What deterioration is normal, what's preventable, and what previous treatment might be judged harshly by future standards? 4. THE TASTE RISK — is this a fashionable period within the maker's output, or a fashionable maker within the category? Distinguish honestly between work considered central to a reputation and work considered peripheral. 5. THE DOCUMENTATION I SHOULD DEMAND NOW — what should I insist on receiving at purchase that will materially help a future sale? Things that are easy to obtain now and nearly impossible to reconstruct later. 6. THE HONEST VERDICT — if I asked you to sell this in fifteen years, would you be pleased to take it, or would you be managing my expectations? Say which. Rules: - You cannot see the object and must not assess authenticity. Work only from what I've described. - Do not predict prices or values. - Where a concern requires a specialist or lawyer to resolve, say so explicitly. Why section 5 is the practical payoff: condition reports, correspondence with the artist’s foundation, exhibition documentation, prior invoices and technical imaging are straightforward to obtain at the point of purchase and frequently impossible to assemble two decades later. Buyers routinely accept whatever paperwork is handed over. Asking for more, at the moment you have maximum leverage, is close to free. Section 6 is the one to sit with. A specialist saying they’d be “managing your expectations” is telling you the piece is a possession, not an asset — which is fine, as long as you’re buying it on that basis. What AI absolutely cannot do here | Never | Why | |---|---| Authenticate or attribute | It cannot see the object, cannot perform technical analysis, and will invent a confident answer. This is the single most dangerous misuse in the category. | Value a work | No reliable access to comparable sales, condition, or current market. Valuations for insurance, probate or sale require a qualified appraiser. | Verify provenance | It can structure the chain and identify gaps you describe. It cannot confirm any link, and must not invent sale records or literature references. | Resolve title questions | Restitution and title are legal matters with jurisdiction-specific rules. A flagged gap goes to a specialist lawyer, not to a chatbot. | Interpret your policy | Insurance wording must be read as written. Use AI to prepare questions for your broker. | Calculate | Model accuracy on financial computation is unreliable. Take the formulas, compute in a spreadsheet. | Which model for which task? Web search · comparables research Never · authentication Use a reasoning-tier model for the exit test and provenance interrogation — both reward careful analysis of how separate weaknesses combine. Use live web search when researching an artist’s market or auction records, and click through to every source: models confidently fabricate sale results and literature citations, which in this field is a specific and well-documented hazard. Take formulas from the cost model and compute them yourself. Auction house fee structures change — Sotheby’s revised premiums in February 2026 — and restitution frameworks are evolving. We re-verify on each review; confirm current fees directly with the house before bidding. Frequently asked questions Is art a good investment? For most private buyers, no, because transaction costs are high enough to consume the returns. Buyer’s premiums at major auction houses run around 26–28% on the first tranche, sellers typically pay commission around 15%, and holding costs include insurance and conservation. Analysis suggests round-trip costs of roughly 15–40% of hammer price, which on a ten-year hold can require appreciation in the region of 55–65% simply to break even. Buy because you want to own it. What are the transaction costs of buying art at auction? The buyer’s premium is the main cost, running around 26% on the first million at major houses and reaching 28% following Sotheby’s February 2026 revision, with reduced rates on higher tranches. Sales tax or VAT is typically charged on the premium as well. When selling, expect seller’s commission plus charges for photography, cataloguing, insurance during consignment and transport. Why does provenance matter so much in art? Because it establishes both authenticity and legal title, and weaknesses in either can render a work unsaleable. Gaps in ownership history between 1933 and 1945 carry particular risk given the scale of wartime confiscation, and even a short unexplained gap in that period can create liability. Restitution mechanisms in several jurisdictions have made claims easier to bring rather than harder, so this risk is not decaying with time. Can AI authenticate a painting or artwork? No, absolutely not — and attempting it is the most dangerous misuse of AI in this field. A language model cannot examine a physical object, has no access to technical analysis, and will produce confident attributions that are simply invented. Authentication requires recognised specialists, technical examination and the relevant artist foundation or catalogue raisonné authority. Use AI only to prepare questions and organise your own research. What are the warning signs of a forged artwork? Pattern-level warning signs include pressure to move quickly, a discovery story that conveniently explains absence from scholarly literature, documentation that can’t be independently verified with named institutions, reluctance to permit technical examination, and a price notably below comparable works. Forged provenance is as significant a risk as a forged object, since fabricated photographs, receipts and ownership histories have deceived experts for extended periods. How should art be insured? Specialist fine art cover on an agreed value basis is generally preferable to inclusion under a general contents policy, which may impose per-item limits well below the value of individual works. Key points to confirm are whether cover is agreed or market value, whether transit and loan are included, how depreciation following damage and restoration is treated, and how frequently valuations must be updated to keep cover valid. What makes a collectible difficult to sell later? Attribution uncertainty, an unexplained provenance gap, condition problems including previous over-restoration, an unfashionable period within an artist’s output, and buying at the top of an artist’s market. Illiquidity is generally discovered at the point of sale rather than at purchase — which is why testing the exit before buying is more useful than estimating future value. Should I buy collectibles through a fund or fractional platform? That’s a financial product decision rather than a collecting decision, and it should be assessed as such with professional advice. Fractional ownership doesn’t remove transaction costs, illiquidity or attribution risk, and it introduces platform, fee and counterparty considerations. It also removes the main non-financial reason to own a passion asset, which is possession and enjoyment. Download: The Passion Asset Acquisition Checklist The pre-purchase checklist, total-cost-of-ownership model with formulas, the provenance question set, pattern-level red flags, the insurance adequacy audit, and the exit test. Enter your email and we’ll send the checklist plus occasional updates. Unsubscribe anytime. Written by the Narracomm team Narracomm is a communications and content strategy team that helps business owners, operators, and founders use AI to produce clear, credible, high-performing work. This guide would benefit from a named reviewer with art market credentials — an accredited appraiser ASA, AAA or RICS , a specialist art lawyer, or a fine art insurance practitioner — given the provenance, title and valuation content. Sources & further reading The Art Newspaper — Sotheby’s adjusts buyer’s premiums and fee structures Feb 2026 https://www.theartnewspaper.com/2026/02/17/sothebys-adjusts-buyers-premiums-fee-structures-securitisation MyArtBroker — Guide to art auction fees https://www.myartbroker.com/auction/articles/guide-to-art-auction-fees Estimonia — Auction house fees compared: buyer’s premiums and commissions 2026 https://estimonia.com/auction-house-fees Sotheby’s — Provenance research and restitution https://www.sothebys.com/en/articles/restitution Claims Conference — Overview of worldwide looted art and provenance research databases https://art.claimscon.org/resources/overview-of-worldwide-looted-art-and-provenance-research-databases/ Withers — Art restitution: ADR mechanisms in Nazi-looted art cases https://www.withersworldwide.com/en-gb/insight/read/art-restitution-adr-mechanisms-to-solve-nazi-looted-art-cases Open Art Data — Provenance gaps 1933–1945 https://www.openartdata.org/2019/11/provenance-gaps-1933-1945-art-DIA.html On one widely-quoted figure: the claim that up to 40% of the art market consists of forgeries originates with art historian Thomas Hoving and is disputed — a number of specialists regard it as substantially overstated. We’ve omitted it from the body of this guide rather than repeat a contested estimate, and note it here only because you’ll encounter it frequently elsewhere. Last reviewed and updated: July 25, 2026 · Fee structures verified on this date; auction houses revise these periodically. Next review due within 14 days. This guide is general information, not investment, legal or valuation advice. It cannot and does not authenticate any work.