Upload the spec sheets a supplier sends and get catalog rows with model number, dimensions, weight, materials, finish options, certifications, and lead time already filled in. Building a product list stops being a week of copy and paste from twenty PDFs.
The app embedded on this page is the CSV Import Tool, a real Taskade Genesis build you can load rows into right now. A product catalog is a wide table, and this is the closest starting point for one. Click "Use this app" to clone it in about ten seconds, define the specification columns below, then add the PDF upload form and the reading step that fills them.
The build turns supplier documents into comparable products. A Products Project holds one row per item with the attributes you actually sell on, and a Suppliers Project sits alongside it as connected records. A published form takes the supplier, the product range, and a File field that accepts several sheets. The Convert File to Text step reads each one, which matters because spec sheets are frequently scanned pages from an older catalog rather than clean exports. An Ask Agent step returns manufacturer, model or SKU, product name, dimensions with units, weight, materials, finish and color options, power or capacity ratings, certifications, warranty, and lead time. Units are requested explicitly, because a catalog where some rows are millimetres and others are inches is worse than no catalog.
The catalog reads in the shapes commerce needs:
- Table view for the full specification grid, filterable by attribute
- Board view to move products through Draft, Reviewed, and Published
- List view for a clean readable sheet per range
- Plus Calendar, Mind Map, Gantt, and Org Chart across the 7 project views
Automations keep the catalog honest. With Taskade automations, a new spec sheet for an existing model flags the attributes that changed, a product missing a required attribute is held out of Published, and an updated lead time notifies sales. Across 100+ bidirectional integrations, Gmail pulls supplier emails in while Google Sheets and Slack push catalog updates out. The AI agents inside the app carry 34 built-in tools with persistent memory, so an agent can write consistent product descriptions from the attributes rather than from a blank page.
Decide the attribute schema before you load a single sheet. A catalog where dimensions live in one column for some products and three for others cannot be filtered, and filtering is the entire reason to build it. Write down which attributes are required for a product to be published, then let the app enforce that rather than a person. Load one supplier range end to end first, fix the schema against what real sheets contain, and only then load the rest. That order saves a rebuild, because supplier documents always contain an attribute you did not think of.
Clone the app, invite merchandising and sales, and the catalog, the source sheets, and the attribute rules are yours.
Read the reading step in Convert File to Text, the pattern in Build a Document Intake App, and structuring in Structured Output. Shape the attribute columns with Custom Fields, browse live builds in the Community Gallery, or explore the templates library. Pair this with Log Deliveries From the Packing Slip in the Box so received stock matches the catalog, and Compare Subcontractor Quotes Side by Side when you are specifying rather than selling.
Frequently Asked Questions
Can it keep units consistent across suppliers?
Ask for the number and its unit as separate values, then convert in a later step to your catalog standard. That is far safer than asking the extraction to convert, because you can see both the source value and the converted one.
What about specification tables inside the PDF?
Tables read well because they are structured on the page. Ask for the attributes by name rather than for the table as a block, and each one comes back as its own field ready for a column.
Can it update an existing product rather than duplicating it?
Yes. Match on manufacturer plus model or SKU, and when a match is found write a change record showing which attributes moved rather than silently overwriting. Merchandising then reviews a short diff instead of a whole sheet.
Does it handle scanned sheets from an old catalog?
Yes. Image-only PDFs and photographed pages are read off the image. A modern typed PDF is read straight through as text with no extra credit cost for the reading.
Can an agent write the product copy too?
It can draft from the extracted attributes, which keeps the copy tied to real specifications. Keep extraction and copywriting as separate steps so you always know which text came off the supplier sheet.
Are the supplier sheets retained?
Yes. Each file attaches to its product row and lands in your workspace Media tab, so a customer question about a rating is answered from the manufacturer document.
What size limits apply?
Scanned sheets should stay under 20 MB each, the ceiling for reading words off an image. Typed PDFs and spreadsheets are read as text and bounded by your plan upload size.
What about products with variants such as sizes or finishes?
Keep variants as their own connected records under a parent product, so a chair in four finishes is one product with four variants. Flattening variants into the product row is the decision people most often regret.
