Create AI Chatbot from Tabular Data using VectorShift AI Agent

Key Takeaways

this is amazing now we going to see Vector shift CSV chatbot in this we are going to create a chatbot like this where you can ask a question like this what is the price of Ford Focus 2018 and then click Send then the chatbot is going to read the CSV which you can see here Ford Focus the pric is 14,000 and it got the information from that and it's responding to the user so how you can create such a chatbot and publish it on your website that's exactly what we're going to see today let's get [Music] started hi everyone I'm really excited to show you about Vector shift CSV chatbot in this when the user is asking a question to the chatbot it is going to use the information from CSV and then respond to the user accordingly if the user ask any question other than the data available in the CSV the chatbot is going to decide and it will use a knowledge base to retrieve relevant information and answer the question so let's take an example Ford Company chatbot so the user is asking question what is the price of Ford Focus and the chatbot is going to read from CSV and respond if the user is asking what is Ford as a company then the chatbot is going to use the knowledge base and then respond accordingly in this we are going to see how to upload data in regards to CSV chatbot the data could be your own CSV or even your own knowledge base such as PDF URL Etc next next we are setting up the pipeline publishing the chatbot and finally deploying on the website so in Vector panel the first step we are going to upload data such as the CSV file with pricing then set up a knowledge base with four data next we are going to create a pipeline which looks like this I'm going to break that down for you and finally publish it on your website like this I'm going to take you through step by step but before that I regularly create videos in regards to Artificial Intelligence on my YouTube channel so do subscribe and click the Bell icon to stay tuned make sure you click the like button so this video can be helpful for many others like here thanks to Vector shift for sponsoring this video so as you probably know Vector shift is no code AI automation platform where you are able to build pipelines automate the whole workflow integrate this with multiple tools and finally publish it on your website I've already covered basics in regards to Vector shift which I will link that in the description below you can get started for free then upgrade if you want to use mvin prais as coupon code to get 20% off now first step we're going to upload data so in your dashboard go to storage first in the file section you can click new upload file and choose the CSV to upload I've uploaded Ford prices. CSV file next go to knowledge bases there you can click new create a knowledge base I'm going to name it as my for you can go to advaned settings if you want to change the chunk size and the embedding model which you want to use I'm going to use text embedding for now and click confirm you can leave it to default it should still work now after coming inside the knowledge base click add documents I'm going to use a URL so I'm going to add this Wikipedia page about Ford company as a knowledge base so I've copied the URL and I'm adding that here you can even upload your own file such as PDF file and many other type of file and click add now in this case I have added URL so now now it start embedding now you can see totally 160 chunks of data taken from the Wikipedia article so what happened here we uploaded data that is a URL then it got divided into chunks in our case is6 chunks and then it got converted to embeddings and it got stored in the database so Vector shift does all the things in the background now next step we're going to create a pipeline and make the user ask a question to the chatbot so coming to our dashboard we are going to the pipeline section you can create a new pipeline from scratch or from template for now I've already created this pipeline which I'm going to explain to you so in simple terms this is exactly how it's going to look like but let's go into more detail so first there's an input field where the user is asking a question then that question is sent to the chatbot or the lodge language model in this case we are using GPD 3.5 turbo you can even choose any other model of your choice and we got various other large language model here next we are providing the information here you are an expert classification agent for fold your job is to classify whether the question is about pricing or about general information so the chat bot is going to decide if the question is about pricing or about Ford company so if it's about pricing then it need to go to the CSV so in instead of directly going to the CSV we are doing another request to the lar language model so this is used to convert a normal instruction to a SQL instruction your job is to convert a question into a specific format the table has three columns model year and price so the CSV which we uploaded contains model year and price so we telling the Lost language model it contains model ear and price column So based on the question the L language model can create a SQL query and run that against the CSV query loader this is a post database so as soon as you upload a file or choose a file which is for prices. tsv this will get converted and loaded in the postgress SQL database then based on the SQL query the output got generated after running the query and finally it is merged and send to the output that's it so that is when the user ask a question about pricing but what if the user is asking about company for then it is going to a different route similar to the one which I've shown above but in this case we are loading the data from a knowledge base so that is my for data that is the URL Wikipedia which we have uploaded which contains 116 chunks next that information is again passed to the large language model to generate a response in this we are providing three different things one is the question next the context from the knowledge reader and third the chat memory which contains the history of the previous chat so question context of the key things which we are pausing here then the large language model is generating the response based on the information provided and finally it's sent to the merch and printed as output that's it so in simple terms we are using a lodge language model at each of this notes to process the request this log language model will check if it needs to go through this path or this path then the laun language model present in CSV and in the knowledge base will extract the data and finally send the output to the user now we got all the pipeline ready we are going to test it by clicking the play button at the top here I'm going to ask a question what is the price of Ford Focus 2018 model and click run and here is the answer the price of a car model with focus and year 2018 is 14,000 that is correct which you can see from this table if you get any errors in this area make sure you go to the upload file section and then re-upload the file directly from here now we are ready to deploy so let's click at the top deploy changes and pipeline deployed next we are going to click the deploy ass as chatbot and provide the relevant information such as the name and the description then click save you can modify the functionality as per your requirement and finally you can export you can click open chat bot and it'll be opened as a interface you can even ask questions directly from here what is the price of for Focus 2018 and you will get a response like this similarly to publish this on website just copy this embed code in my case I'm going to use my WordPress website go to appearance and editor there I'm going to click templates and then page with sidebar and click fter then add the embedding code there next save and going to view the page now I can see the chatboard on my right hand side I can ask question directly here what is the pricee of Ford Focus 2018 and you got the response here this is exciting now you are able to create a chatbot and publish it on your website with your CSV data like this you can even integrate that with multiple different data source providers such as air table Google Docs Google Drive Google Sheets and much more this will take your application to the next level I'm going to create more videos similar to this so stay tuned I hope you like this video do like share and subscribe and thanks for watch watching