Adobe PDF Services StrawberryRenewal dates out of forty supplier agreements
Including the ones that arrived as photographs of pages. A companion fetches each file, runs the conversion, and reads all forty.
Some documents are text pretending to be an image. A supplier sends a signed agreement as a scan, a council publishes a planning decision as photographs of pages, a service charge statement arrives as a fax that was printed and scanned twice. There is nothing to copy and nothing to search.
A companion fetches the file from wherever it sits, runs the conversion through an account you connect, and then does the job the document was for: the renewal dates, the purchase order references, the clauses that mention termination. Forty documents, one pass.
When is a conversion step necessary?
When the PDF has no text layer.
Try to select a sentence. If nothing highlights, you have an image, and any tool claiming to have read it has either guessed or run OCR of its own. For a one-page invoice a guess is often fine. For a 94-page contract where a clause number matters, it is not.
The other case is structure. A table in a PDF is drawn lines and floating text with no relationship between them. Extraction that recovers rows and columns is a different operation from reading the words, and it decides whether you get a number you can total or a number you retype.
What is the practical workflow?
Convert, read, then do the work. A companion runs the conversion, gets text or structured output back, and uses it for the point of the exercise: renewal dates out of forty supplier agreements, which invoices reference a purchase order, the clauses that mention termination. For contracts specifically, ai for legal ops covers who owns the reading.
The Available tools list on the Adobe PDF Services entry under `strawberry://settings/integrations` shows exactly which conversions and extractions your connection exposes, with read-only tools marked. Each conversion is a metered call against your Adobe allowance, so a batch has a cost attached.
- Test whether the PDF has a text layer before assuming a conversion is needed.
- Structured table extraction is a different operation from reading the text.
- Each conversion is a metered API call, so batch work has a cost attached.
- The document goes to Adobe for processing, which is a data-handling decision worth making deliberately.
Getting the document there is the other half
The file is usually sitting in an email, a portal you have to log into, or a shared drive, and fetching it is the part that involves clicking. A companion handles that half: signing into the supplier portal, downloading the statement, and taking it from there.
Fetching, converting and reading happen in one pass, for all forty documents. The recurring version is close to warranty and receipt filing or purchase order matching, and both stall on whether anybody opens forty PDFs.
Experience Strawberry for free
DownloadTrusted by fast-growing companies worldwide
Frequently asked questions
Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed · Canonical facts for AI agents