Google Drive “Upload File” fails after Merge, even though binary data is present

Questions
1

I’m downloading a PDF attachment from Outlook and only want to upload it to Google Drive after a GPT-based check confirms it’s an invoice. To do that, I:

  1. Download the attachment (binary.data contains the PDF)
  2. Extract text and send to LLM, which returns isInvoice: "ja" or "nein"
  3. Branch on isInvoice using an If‑node
  4. Merge the two branches (Append mode) so I have both the JSON result and the original binary in one item
  5. Feed the merged output into the Google Drive Upload File node

In the Merge node’s output preview I can see under the Binary tab that data holds my PDF. However, the Upload File node still throws:

This operation expects the node’s input data to contain a binary file 'data', but none was found [item 0]

even though the Merge preview clearly shows binary.data.

grafik
grafik1658×576 65.6 KB

grafik
grafik1844×916 73.4 KB

2

Could you please share the workflow itself instead of the screenshot. This is how.

I also see that there are two AI blocks in the workflow. One is checking for whether the pdf is an invoice. What the second is for?

3
4

I also see that there are 4 inputs to Upload node

image
image766×740 53.6 KB

are they 4 binaries?

5

Hi no there only should be one data bc the input is only one pdf

6

So there are inputs, where there is NO binary (if only one has binary). This will cause the upload node to be not happy, right?

7

if my gmail gets a email with a attchemtn a file the workflow starts and get a binary data the pdf that was in the mail

8

If you switch to JSON, Schema or Table, what are other inputs?

9

hey what do you mean?

10

Here, if you switch from the Binary tab to other tabs

image
image583×476 19.2 KB

what you see there - do you expect to see there?

11

from there i only need mz text that is the id for the right folder where the file should be uploaded but this is ecerything right and also the data is there but idk why it dont uplaod the data and says that there is no data?

12

OK, looking at workflow, the first AI agent takes the text from the pdf and returns the company name, the date and whether it is an invoice.

then

  • if it isn’t an invoice - it bypasses the second AI Agent and goes to the upload node (is that intended)?
  • if it is an invoice - it goes to the GDrive search node which searches for “Invoice” in the name of folders and files. How is this information used later? I do not see this node used. After this search is done, the flow goes into the second AI Agent, where the prompt takes the date from the first AI agent and you are asking it something like

Now use the invoice date to find the matching ID and output only that ID (without quotes or any dashes — just the ID).

^ what does this mean, what should the agent match the date from the first agent to?