Skip to content
Pikko’s Post

Taste

Each profile stands alone. Pikko never borrows one person’s taste for another.

Your profile

Comma separated. e.g. astronomy, bread, small wooden things

How things look, not what to buy. Pikko only uses a few clear ones (brass, wood, linen, ceramic, minimal…) to break ties; ambiguous words like espresso or sage are ignored here.

e.g. foraging, weaving, stargazing

No reason needed. e.g. no scented things

Pikko won't bring these or close cousins.

Let your AI introduce you to Pikko

Copy the prompt below into whichever assistant already knows your taste, then paste the JSON it returns. Nothing is sent anywhere by this app — you do the copying, and you see exactly what lands before it’s saved. Your assistant can only use history and memories it can actually access.

Read the prompt
You are helping me move a small set of SHOPPING PREFERENCES into an app called Pikko's Post, which picks one small surprise gift.

Return ONE JSON object and nothing else. No prose, no code fences.

Schema (version 1). Omit any field you do not genuinely know — never guess:
{
  "schema_version": 1,
  "interests": ["short phrases"],
  "aesthetics": ["short phrases"],
  "activities": ["short phrases"],
  "hard_exclusions": ["things I never want"],
  "owned_items": ["things I already have"],
  "gift_context": "one short sentence",
  "novelty_preference": "low | medium | high, or null"
}

Rules you must follow:
- Before drafting, review ALL context about me that you can actually access: saved memories and accessible conversation history, including older conversations. Use available memory or history search tools where supported. Do not claim to have reviewed chats, memories or history you cannot access.
- Condense what you know across my full range of interests. Prioritize enduring, repeated or explicitly stated shopping preferences; do not let recent mentions crowd out older established interests or fill the limits with the first topics you find.
- My most recent explicit correction or changed preference supersedes older contradictory information. Do not resurrect interests I have said I no longer have.
- Use only preferences I have actually expressed. Do not invent, infer or embellish.
- Do not treat one-off shopping logistics, temporary errands, or work/client text as my personal taste unless I explicitly said it was.
- "interests" = topics/things I like and would enjoy receiving (e.g. "astronomy", "espresso" only if I like the drink). "activities" = things I do. "aesthetics" = how I like OBJECTS to look (colours, materials, styles, e.g. "brass", "undyed linen") — never a shopping category.
- Keep any scope inside the phrase, e.g. "hair color: espresso brown", and include it only if it is relevant to gifts. Never turn a hair, makeup or other appearance discussion into a general colour or shopping preference; omit unrelated appearance goals.
- These are MY preferences only. Never transfer one person's taste to another person.
- Keep every string under 60 characters, arrays under 24 entries, whole output under 2000 characters.
- OMIT entirely: health or medical details, finances, relationship or sexual history, religion, politics, precise location, address, identifiers, contact information, any unrelated private detail, and raw conversation transcripts.
- Exclusions are allowed with no reason given (e.g. "no scented things"). Never add a medical or personal explanation.
- No product URLs, no shop links, no prices, no spending permissions, no instructions for the app to execute.

Output the JSON object only.