Your Catalog Has 120 Products. You Shouldn't Have to Tag Them One by One.

The setup work for a digital catalog is hidden. Buyers see a clean flipbook with clickable products. What they don't see is the hour or two a seller spent before that — opening the catalog on one screen, clicking a pin tool, dragging a hotspot to a product, typing a name, a price, a SKU, moving to the next product, and repeating this for every item across every page. On a 30-page seasonal catalog with four products per spread, that's 120 individual operations before a single buyer has seen anything.

It's not hard work in the traditional sense. It's the kind of repetitive, detail-heavy work that's easy to do wrong when you're tired, and impossible to scale when your catalog grows. And it's also exactly the kind of work that AI handles better than people do.

What AI product detection actually does

When you open the Product Pins editor in Pholio and click "Auto-detect this page," the AI reads the page the way a person would — visually, as an image — and returns a list of products it found: name, price, position on the page, SKU if visible. For most catalog pages, this takes a few seconds.

The result isn't placed immediately. It comes back as an editable list — you can rename a product the AI misread, fix a price, delete a false positive, or add something it missed. Nothing is written to your catalog until you press Confirm. The AI does the first pass; you do the last check.

The AI does the first pass. You do the last check. For most catalog pages, that last check takes about thirty seconds.

This matters for a specific reason. A seller knows their own products. They know that "CUP CLOVE 18cl" in the catalog is actually listed in their system as "Cup Clove 180ml," and they know which price is the current season's wholesale rate versus a sample price that snuck in from a production draft. The AI gets close; the seller gets it right. That division of labour is what makes the feature actually useful rather than just technically impressive.

Scanning a whole catalog in one pass

For a seasonal catalog launch — 40 pages, fresh from the designer, every product new — doing one page at a time is still faster than doing it manually. But Pholio also has a whole-catalog scan: it runs through every page in sequence, tags everything it finds, and bulk-creates all the pins at once. The upfront cost is two AI credits per page, charged as each page is processed. If you run out of credits mid-catalog, the scan stops and keeps whatever it found — you can top up credits and continue.

The typical use case is a new season upload. A seller drops in the PDF, hits Scan whole catalog, waits a minute or two, reviews the output, corrects anything off, and confirms. The catalog is live with products tagged and clickable, in a fraction of the time it would take to tag it manually.

When there's no PDF — building from a spreadsheet

Not every brand has a designed PDF ready. Some run their catalog from an Excel sheet or a price list, especially for early-season previews or smaller collections. Pholio's Quick Catalog feature turns a CSV or Excel file into a shareable digital catalog directly — no PDF needed.

The AI step here is column mapping. Spreadsheets rarely have consistent column names: one might say "Wholesale Price," another "W/Sale," another "Price (EUR)." Pholio reads the header row, figures out which column maps to which product field — name, SKU, wholesale price, RRP, description, size — and builds the catalog automatically. One AI credit per import. If you're on a plan without credits, it falls back to a free heuristic pass that handles the most common column patterns.

The result is the same buyer-facing experience as a PDF flipbook: a shareable link, a product grid, an inquiry panel on the right. Buyers can wishlist items and submit an inquiry exactly the same way. The only difference is the seller started from a spreadsheet instead of a designed file.

AI Insights: which products are actually selling

Tagging products is the setup. The question that comes after is: are the right products getting attention?

Pholio's Top Products section shows a ranked list of best performers across orders, inquiries, and wishlist-only adds — a product that's frequently wishlisted but rarely ordered is a different problem than a product that's rarely seen at all. Below the table, the AI Insights panel reads this data and generates strategic commentary: which products are outperforming relative to inquiry count, which buyer countries are under-indexed, where price might be holding back conversion.

This runs on one AI credit per generation. There's no cooldown — you can regenerate as often as you have credits, and you can add a focus prompt to direct the analysis: "Focus on our autumn collection" or "Which products have room to raise the price?" The analysis comes back as a short list of bullets, categorised as positive signals, warnings, or suggestions.

It's not a replacement for knowing your own business. It's a second opinion from something that has read every order and inquiry you've received and has no stake in the answer.

How credits work

All of Pholio's AI features run on a single credit balance. Trade plan includes 50 credits per month; Studio plan includes 200. Credits reset on the first of each month. Non-expiring top-up packs are available — 50 credits for €5, 200 for €15 — and unused balance carries over indefinitely.

One credit is charged for: an AI page scan, a whole-catalog scan (2 per page), an AI Insights generation, an AI Quick Catalog import from a spreadsheet, or an AI sheet import into an existing PDF catalog's pins. The credit is only consumed when the AI call actually runs — if credits run out, the feature either blocks gracefully or falls back to a free path, depending on the feature.

The logic behind the flat-per-action pricing (rather than per-product or per-page): the expensive part is the AI call, not the number of items returned. A page with eight products costs the same to scan as a page with two. The credit system reflects that.

What this looks like in practice

A wholesale brand uploads their new season PDF on a Monday morning. They scan the whole catalog — 48 pages, two AI credits per page, 96 credits total. The scan takes about three minutes. They spend another ten reviewing the output, renaming two products the AI got slightly wrong, deleting one false positive on a decorative spread. They confirm, and the catalog is live.

That afternoon, they share the link with their buyer list. Over the next week, inquiries come in. By Friday, the AI Insights panel tells them that their ceramic collection is getting significantly more wishlist adds than orders relative to the rest of the range — which usually means the price point is right but buyers want to see it in person first. They know to follow up on those wishlists specifically before the next trade show.

That's the full loop: AI handles the setup, the data accumulates automatically, and AI reads the data back as something actionable. The seller stays in the parts that require judgement. The parts that don't, don't require them.