COLLECTOR GUIDES
Discover your next bottle with ChatGPT
Your collection can help you choose what to explore next. Ask ChatGPT to look for patterns in your own tasting history, or examine the variety on your shelf, then build a researched shortlist for your wishlist.
Start with one question: would you like more of what you already enjoy, or something different? The two approaches use the same cabinet but lead to different suggestions. Recommendations happen in the conversation; only the wishes you choose are saved.
1. Read the evidence before recommending
Enable your My Bottle Collection connection in ChatGPT and name the cabinet. Collection reads use your existing access; saving wishlist entries needs Owner or Editor access and write permission. Web research uses ChatGPT’s browsing capability separately from the collection connection, when available.
“Read my [cabinet name] collection and wishlist. Include finished bottles when reviewing tasting history. Read the relevant bottle details, my tasting notes and my own ratings where available. Keep other people’s ratings separate. Tell me which records you reviewed, what is missing and whether any pages remain unread. Do not change anything or recommend bottles yet.”
Ask ChatGPT to follow additional result pages before claiming to cover the whole cabinet. A large collection may need several passes. Each physical bottle is its own record: distinguish the number of bottles from the number of distinct expressions, so several bottles of one favourite do not look like several different styles.
Owning a bottle does not prove you like it. An unrated bottle is not a dislike, and an unrecorded tasting is not evidence that you have never tried it. Shared pour notes and guest reviews may describe other people’s tastes. Ask ChatGPT to identify whose preferences it is using and to leave uncertain attribution unresolved.
2. Follow the flavours you enjoy
Use your own notes and ratings as the starting point, then tell ChatGPT about preferences you have not recorded. Useful constraints include category, country of purchase, budget in a named currency, flavours to avoid and whether you want an everyday bottle or something for an occasion.
“Using my own ratings and clearly attributable tasting notes, describe the preferences supported by my history. Cite the bottle names and IDs behind each pattern and separate evidence from guesses. Ask about gaps. Then propose three bottles I might enjoy for up to AUD 150 each, available to buy in Australia. Exclude expressions already in my collection or wishlist unless you explain why a repeat is useful. Do not save anything.”
If there are few personal notes or ratings, ask for a short set of questions instead of a confident taste profile. “I enjoy these two bottles but dislike heavy smoke” can be a better starting point than purchase counts. The tasting journal and ratings guide explain how to build that history.
3. Choose how you want to broaden the collection
For a different approach, ask ChatGPT to describe what is represented and suggest contrasts you would enjoy exploring. You can examine the bottles currently on the shelf, the full collection including finished bottles, or both as separate views.
“Analyse the variety among my current bottles by category, recorded style, producer and ABV. Count physical bottles and distinct expressions separately. For region, cask type or flavour profile, distinguish recorded facts from sourced research and unknowns. Show which areas are well represented, then suggest three contrasting options within AUD 150 each. Explain what each would add and ask which direction interests me. Do not add wishes yet.”
Region, cask and flavour are not dedicated structured fields in the current cabinet. Use them only when supported by label details, notes or clearly identified research. Do not fill gaps from a producer’s reputation or imply that every expression shares the same character.
Variety is a choice, not a score you need to maximise. A cabinet focused on one region or style may be exactly what you want. Ask for a nearby step outside your favourites or a larger contrast, rather than treating every unrepresented category as a shopping requirement.
4. Turn the ideas into a researched shortlist
Ask ChatGPT to verify the exact expression and bottle size. Producer pages are useful for product details; retailer pages can establish a listed price and availability in your country. Prices and stock can change, so ask for source links and the date checked. If browsing is unavailable, keep the ideas as unverified suggestions and check them yourself.
“Research the shortlisted expressions. Show official product sources where available and current retailer links for my country, with the date checked. Compare exact edition, size, ABV, listed price and currency, why it suits my preferences or adds variety, and any uncertainty. Distinguish delivery charges and unavailable stock. Keep my private notes and reviewer details out of web searches. Do not buy anything or change the cabinet.”
| Column | What it helps you decide |
|---|---|
| Exact bottle and size | Whether the research matches the expression and format you want. |
| Why this bottle | Which personal preference or chosen contrast supports the suggestion. |
| Already owned or wished for? | Whether this is a new expression, a deliberate repeat or an existing wish. |
| Sources and date checked | Where product details, listed price and availability came from. |
| Price and open questions | Currency, delivery, budget fit and details still to verify. |
A recommendation is an informed suggestion, not a guarantee you will enjoy the bottle. Keep your own rating, the cabinet average and an external review distinct. Price research is not a valuation of your collection.
5. Add only the bottles you choose to the wishlist
Pick your favourites from the shortlist. Ask ChatGPT to read the current wishlist again, check for existing entries and show the proposed name, HTTPS reference link, notes and optional target price. The target is what you are willing to pay in the cabinet currency; it is separate from a retailer’s current asking price.
“I choose shortlist items 1 and 3 for my [cabinet name] wishlist. Read the latest wishlist and check for existing entries first. Show the exact names, reference links and a short reason for each choice. Leave target prices blank unless I give them. Do not create stock or change my existing wishes. Show me the proposed entries before saving.”
Once the preview is right, give the final instruction:
“Add those two reviewed entries to my wishlist with the agreed links and notes. Leave their target prices unknown. Return the saved wish IDs and read back the wishlist so I can check them. If one already exists, report it rather than adding another.”
Wishlist entries are added individually. If only one succeeds, retain its ID and check the latest list before continuing. If a result is uncertain, ask ChatGPT to check what was saved before retrying. Keep the original request identity for a retry rather than issuing a fresh addition.
Check the Wishlist tab on the website when finished. No purchase or stock change is needed for this workflow. When a bottle later arrives, use From wishlist to shelf to record the actual quantity and amount paid.