The brief, the read, the shortlist.
Your Personal Shopper is not an algorithm that guesses what you might want. It works from the brief you write, reads every new listing against it, and brings you a short, considered shortlist. Below, how each step actually works.
You tell it what matters.
A brief is not a search query. It is a standing instruction. Taste tags, colours, fits, houses you trust, houses you do not. Budget ceiling. Tone. Occasion.
Briefs can be a list of rules, or written in sentences in chat, and learn from the pieces you return to.
It reads the agent sheet.
Under every listing sits an agent-readable product sheet: brand context, condition, fit notes, styling cues, buyer-intent signals. Humans rarely see it. Your Personal Shopper always does.
This is what makes the shortlist different from an algorithm. The brand has already written the answer. Your Personal Shopper just has to read.
Architectural seam-work, cotton poplin, worn twice. Day-to-evening. Runs small, size up if between. Provenance chain intact.
It brings a few, not many.
The work is in what it leaves out. A considered shortlist is three to six pieces, not thirty. Each comes with a short explanation, written to be scanned, not read.
By email: a weekly digest. In chat: a thread you can argue with.
Quiet by default.
It will not guess at wishes. If your brief is silent on jewellery, jewellery stays out of the shortlist.
It will not surface pieces to meet a quota. The shortlist can be empty. Silence is a valid answer.
Your brief is yours. It is not pooled, not sold, not used to train a global model.
Auto-bidding only triggers inside the budget ceiling and rules you set. Approvals, not recommendations.