You have to open the data source and extract the data to feed into the LLM. For the PDF, if the document is machine generated (read: not scanned), you can use the PDF: Extract text action to get the raw text. For the DOCX file, you might have to use a special package or automate opening the file in Word, selecting all the text, putting it on the clipboard, then copying the clipboard into a variable. ​@subbu2025 does that make sense?

Thanks ​@Aaron.Gleason. I am trying to implement Agentic AI for data extraction, as the pdf document contains around 300 pages and need to extract few of the content by using AI > Skill ( Prompt).  Currently from  AI > Skill able to provide input source using variable, my question is how to reference to a document (either word/pdf) instead of providing source string. Hope its clear. Please let me know if any queries. 

Hi ​@subbu2025 currently, the AI Skills use string variables to accept inputs. For intelligent document processing of PDFs, you might want to try Document Automation.

​@Shreya.Kumar ​@subbu2025 Correct, as always! Another possibility is to use the new https://ai.automationanywhere.com/ and upload the document to a knowledge base. Be aware that it takes some time before the documents are queryable, but once they are processed, you have a full RAG* to query. 

* retrieval augmented generation, essentially an LLM trained on your own documents

Sure Thanks ​@Aaron.Gleason ​@Shreya.Kumar for your inputs. 

Hi ​, I have below queries. 

1. https://ai.automationanywhere.com/  -  Do we have any documentation link to refer and whether this is available for use case implementation for client. Please confirm.
2. Whether we can integrate with the task bot as well. Basically the Agent once its created using above URL.

Also, as per my requirement, I need to use the above URL and include the document in Knowledge base and from there create workflow to fetch the content from it. How to set the trigger condition for this to trigger the workflow?

Thanks for your time.

​Hi @subbu2025

1. Here is the documentation link: https://ai-kb.automationanywhere.com/docs/projects/
2. In order to integrate the AI Agents from above URL to our Task Bots, you can use AAI Enterprise Knowledge Package. More info on that in below link: https://docs.automationanywhere.com/bundle/enterprise-v2019/page/gen-ai-studio-aa-enterprise-knowledge.html

Thanks ​@SejalDhondkar for your prompt response.

One more query, using https://ai.automationanywhere.com/  I have registered and once signed in I am able to see the options but post creation of project the project is not listed post refresh looks empty(Below is the screenshot for reference)

2. Do we need to add any specific roles and permissions. Also unable to save the custom agents post creation. Please assist.

​Hi @subbu2025 

Check if you are seeing below rights in your project settings and Role as Admin in your Account Settings. These roles are provided by default hence you should not need to take any steps manually.

Also check once if cookies are enabled.

​@subbu2025 As per my knowledge, the default roles given after signin are enough to create projects and Agents. 

Are you still facing the issue? 

hi ​@SejalDhondkar  - yes, post refresh unable to see the created projected. No projects are getting displayed. I have noticed my Account its showing as “Signed in as Anonymous”. any idea on this?

2. Trying to create new Agent and unable to save the agent. Getting error as “Error saving agent . please try again”.
3. How to see all the created projects and create an agent for an poc. 

Please assist on this.

Hi ​@subbu2025 

You should be able to see your logged in email ID in account settings, and have admin rights by default.

Do check if cookies are enabled once for the site since it seems a authorisation related issue.