- Print
- DarkLight
- PDF
Introducing the chat tab
Chat is the main tab of Paradigm. It's here that you can have a conversation with your private AI assistant.
Remember that the top right box Search through the docs should be checked upon creation of the chat session if you want your assistant to look for answers based on your documents.
Session history
Each conversation that you have with your assistant is called a chat session. You can see the history of your sessions in the left bar. We recommend you open a new fresh session by clicking "new chat" when you switch to another discussion topic or want your assistant to start with a fresh context.
Documents
Accessible by clicking on the corresponding bottom left icon, the documents tab is the place where you can see all the documents your assistant has access to answer your questions.
The export button exports the list of documents present in your collection. The search bar enables you to search for document titles and check whether a document is in your collections.
Click on the top right upload button to add documents to your collection and expand the knowledge your assistant has access to.
Click on the three vertical dots in the row corresponding to a document, and click delete.
Note: Uploading and/or deleting documents might require the corresponding permissions.
Evaluation and feedback
When satistified or dissatisfied with a response from your assistant, we encourage you to click the like/dislike button to provide feedback. This will help your administrator to monitor user satisfaction and might be useful to further improve the assistant.
Feedback data is only stored on the infrastructure where Paradigm is located. This means that if Paradigm is deployed on-prem / on your private cloud, only Paradigm administrators from your company can access this data.
Regenerate button
Your assistant is powered by an LLM which by construction is a stochastic algorithm. This button lets you reroll the dice and get a different answer for your current question.
Specifying Documents for Search in Chat
The document specification feature for search in chat allows users to select the documents they want to search within when asking a question. This helps to narrow down the search scope and get more accurate results.
To activate the feature, click on the magnifying glass icon located in the chat input field. This will open a modal window that allows you to select the desired documents.
You can select documents in two ways:
- By selecting from all documents within your organization or workspaces you have access to.
- By selecting documents from your personal space.
You can search for documents using a dedicated search field that allows you to search by document name, owner name and workspace name.
You can select all the documents appearing in your current search by ticking the "Select All Documents" option.
Once you've selected document(s), the selected document(s) are added to the chat input field and remain present in the current session until you delete them using the cross icon. This allows you to specify the target documents for the session without having to re-select them.
You can delete selected documents by clicking on the cross icon next to each document.
Upload Documents in Chat
You can directly upload one or multiple document(s) in Chat by using the attachment icon in the entry bar.
It will allow you to upload documents, and will then scope your query on this document(s) once the upload completed.
Documents selection for follow-up questions in Chat
When relevant documents are identified to support an answer, you have the option to select from among those documents which ones you would like to use to continue the conversation in Chat.
You can do so by clicking the grey + sign on a relevant document.
You will then see it (or them) in the chat document selection zone as selected for the rest of your conversation