Catalog AI
Agent · AI Import

Your supplier sent a 300-page PDF. Now it is a catalog.

AI Import reads PDFs, spreadsheets, images, and plain text, and gives you back structured products to review. Approve what you want, and enrich it in the same run.

Import your first file freeBook a demo

Import · vendor-catalog-q3.pdf

41 products extracted · awaiting review

Review
Page 14Angle grinder, 4½ in · AG-450-Atable row 3
Page 14Angle grinder, 5 in · AG-500-Atable row 4
Page 15Cut-off wheel, 4½ in × 0.045 intable row 1
Page 15Flap disc, 40 grit, zirconiatable row 2
The situation

Every new supplier arrives as a retyping job.

One sends a spreadsheet with merged cells. One sends a PDF that was designed to be printed. One sends photographs of a printed price list, which is somehow the most common of the three.

None of it matches your schema, and all of it has to be in the catalog before anything can be sold. So it gets typed, by someone, over a week you did not budget for.

AI Import reads the file the way a person would and hands you rows to approve instead of a document to transcribe.

Formats

What it reads, and how well

Stated honestly, because knowing which files will be awkward is more useful than a logo grid.

CSV · XLSX · XLS
First row as headers, one product per row. Column order does not matter and column names do not have to match yours.
PDF
Best with structured layouts and tables — supplier catalogs, price lists, spec sheets. Scanned pages work, at lower accuracy.
JPG · PNG · WebP
Product labels, packaging shots, and photographed datasheets. Clear, well-lit images extract best.
TXT
Both structured key-value text and unstructured prose. Useful for email-pasted product lists.
Limits
Up to 5 files per import, 20 MB each, 100 MB in total.
Process

How an import runs

  1. STEP 1

    Upload the files

    Drag in whatever the supplier sent. Mixed formats in one import are fine — a PDF catalog and a price spreadsheet can go in together.

  2. STEP 2

    Choose the mode

    Review and import, where nothing is written until you approve it. Or import and enrich, where extraction flows straight into an enrichment job.

  3. STEP 3

    Guide the extraction

    Add a prompt if the file needs one — "extract product name, SKU, price, and available colours" — and pick a model tier.

  4. STEP 4

    Review and commit

    Extraction runs in the background. You get a row-by-row review screen and choose exactly which products enter the catalog.

Modes

Two ways to run it

The difference is where the human sits in the loop.

Careful

Review and import

Extraction finishes, you get a row-by-row review screen, and nothing touches the catalog until you approve it. The right choice for a new supplier or an unfamiliar format.

Fast

Import and enrich

Extracted products are finalised and an enrichment job starts on them automatically. The right choice once you trust the source and want the whole range live by tomorrow.
AI Import

Practical questions

What if the PDF is a scan rather than real text?
It still works, with lower accuracy. Those rows tend to come through with obvious gaps rather than subtle errors, which is why the review step exists — and why running validation afterwards is worth the credits.
Does the spreadsheet have to match our column names?
No. Headers are interpreted rather than matched, so a supplier calling a column "Art.-Nr." and you calling it "SKU" is not a problem worth solving by hand.
Can it read a 300-page catalog?
Within the 20 MB per file and 100 MB total limits, yes. Long documents are the case it was built for — that is where the manual alternative hurts most.
Will it create duplicates?
You see every extracted row before anything is written, and you choose what imports. Nothing enters the catalog behind your back.
Can it enrich in the same run?
Yes — import and enrich finalises the extracted products and starts an enrichment job on them automatically.
What does it cost?
Extraction consumes credits based on the model tier and the size of what it is reading. As with every job, the estimate comes before the run.
The rest of the platform

Six agents, one catalog

Product Enrichment

What is this product?Fill every attribute from live sources.

Product Validation

Is any of this still true?Score the data you already have, 0–100.

Market Analysis

Who else sells it, and for how much?Competitor price, MSRP, and stock status.

Brand Enrichment

Who makes it?Brand profiles written at catalog scale.

Category Enrichment

Where does it belong?Category copy, meta content, and slugs.

Upload the worst file on your desk.

The supplier catalog nobody wants to open is the right first test. Free credits cover the extraction.

Start freeBook a demo

5 credits on signup · no card required