Brain AI
Enhance your customer support with our new GEN-AI platform. Train a custom knowledge base using advanced Large Language Models (LLM) to deliver smarter, more accurate responses to your customers.
Last updated
Enhance your customer support with our new GEN-AI platform. Train a custom knowledge base using advanced Large Language Models (LLM) to deliver smarter, more accurate responses to your customers.
Last updated
Knowledgebase is a centralized repository of information. Using our Gen-Ai platform, you can create & train a custom knowledgebase with your company's product's documentation, privacy policy, warranty & return policy etc. . This knowledgebases help chatbots to deliver smarter, more accurate responses in a more human manner.
In the Data Source sub-tab, users can view all their existing knowledge bases or create a new one. Additionally, users can see the number of tokens available—tokens act as a digital currency needed to create and train a knowledge base in REVE Chat. For more details on tokens, refer to the pricing section.
The Insight Tab in our Knowledge Base feature provides a comprehensive view of all reports and analytics related to your knowledge base's performance. This tab is your go-to resource for understanding how many tokens used for training & query response, how well your knowledge base is functioning using response performance.
For a new knowledge base click the "Create Knowledge Base" button. There are two types of large language model available for training. One is LLaMa2 and the other one is Chat GPT.
A pop-up will appear where you have to provide a name to your knowledge base. After providing a name, you will land in the data source page of that knowledge base.
In the data source page, we will add our training documents by clicking the Add Data Source button. A pop-up will appear where we have to select which data type we want to use
For Website URL, you can add your company URL, policy URL or product documentation URL. And if you enable the Fetch Link check box, all existing web link under the main URL will be fetched automatically. You can limit the fetched URL links by providing how many do you want(available when fetch link checkbox is marked).
For Manual URL, you can add each web link manually by simply pasting one after another, either side by side or under the each other. It is mainly useful, if you have changed some content of a specific URL and want to update in the knowledge base as well.
For Add Document, you can add electric document files which you want to add to your knowledge base.
After selecting your source knowledge base file, click add data. It takes couple of seconds to minutes based on number & size. During this content upload period, you can not add any new content. Only after completing the upload, users can upload again.
After adding your specific URL or document, click the Train Data button to start the process. To complete the training, it does require some time based on shared URL or file size. Your knowledge base is ready for use.
After training your chatbots with different materials, now it is time to implement it in chatbot. We have brought a new Gen Ai action using which businesses can reply to any open-ended queries (related to the training material).
To use generetive ai, you have to add the new Gen Ai action in a node you think will best serve your customers. After adding the action you have to-
Select a knowledge base from the drop down(green box). Only those knowledge bases which are trained can be selected.
Next you can limit the number of query response by adding a number. After chatbot provides that many knowledge base generated response, the flow will continue to other nodes or action
In this action, users can also add quick replies to direct them to different parts of the flow
To access Gen-Ai's chatbot settings, navigate to Settings> Gen Ai Config (green box). Here you can enable visitor feedback option to check if the knowledgebase is able to answer their queries. You can also enable the visitor to view the source so that they could understand that their query statement is not being misinterpreted.
Besides using the Gen Ai action in certain parts of the chatbot, the knowledge base can be used to answer any open-ended queries by changing the response priority(Yellow box). There are 3 options available:
Always Respond From Intent Samples: Here the open ended queries will be answered from the intent sample training, no LLM based response will be given
Always Respond From Knowledgebase: Here all the open ended queries , will be answered using the trained knowledgebase
Intent Sample First Then Respond From Knowledgebase: The bot will first search for a match in the intent samples. If no suitable match is found it will be refered to the knowledgebase to generate a response.
Any questions? Please email us at issue.support@revechat.com or scan to chat with us.