Hi,
I need to analyze scanned pdf with Google Vertex in StudioX - but the 3 available actions do not have a file parameter, to specify what file needs to be analyzed.
Is there a way to that? I need to use Google Vertex.
Thanks,
Gabriela
The built in activities are for Text related and not image…so with default activities you cannot…
go to vertex find the endpoint and you can try hitting from UiPath
cheers
Hi Gabriela,
The short answer is: you can’t. The default activities in that package don’t have an implementation for the document analyser.
I don’t know how extensive is your UiPath/Software development knowledge, but… an alternative could be creating a custom connector (Or building an API you could invoke from UiPath) with the exact scope you need to call from Vertex. Please consider it will add significant complexity to your implementation.
I personally am not so familiarized with Vertex, so sadly this is all the input I could give.
Parse and chunk documents | Vertex AI Agent Builder | Google Cloud
1 Like
I have solved this issue in several ways.
For UiPath StudioX I found a way around it using Handwriting Detection Activity within Google Vision Package (I have configured the Google Vision connector in Integration Service).
For UiPath Studio that works on a OnPrem Orchestrator, where I do not have an Integration Service and connectors available, I have built a library that connects to Vertex through http request. This library also features system prompt and history capabilities.
Having the http request model made for the automation developers I could replicate the same logic on the StudioX.
I haven’t figured out how to make a custom connector yet though, or how I could import a library made on Studio, in StudioX… This is still something I have to figure out…
Clever solution, thanks for letting us know.
The custom connector could be a little tricky, since every API provider is different. In general if you understand the basics of API authentication, you should be able to figure it out.
Try this
- first you will need to create credentials on Google’s side
- depending on the auth method you choose you could either get an API key or an OAuth client
- Configure the key/client accordingly in Google
- Create the connector on UiPath, go to Authentication and fill in the required parameters
- Once you have successfully authenticated in your connector, the rest is creating the methods (get, post, detelte…etc) , for this refer to the documentation of the APIs you want to use from google
