Hello, my name is Alexander. I am an AI automation specialist and I'm created this scenario to help you export your uh database into vector database and it will help you to create AI agents that will use this database and can answer on your question. So I have ready to use team players but I am proposed to start from the scratch. So let's create a new project in NA10 and we need to create a new trigger. Then we need to grab this data. For example, I need a property listing. I just let me remove filters. Yeah, we have like a 100 records in this table. It's just loading. And uh we need to we need to create API request to grab all of these records and put into our NA10. So we have an air table API list. Uh if you're working before you probably see that we have some tools, but it's like base. We have a base and we have a records. But I don't see a list here. For example, I want to grab only this list. Like I don't want to grab all of the base like horizontal estate. So we just need to create like HTTP request. Then we need to import our curl. Copy. For example, we need to find a list record. Click on copy. put this comment here. Uh we need our URL. How you can get this URL? It's include like a base ID and table or name ID. So first part base ID you can see in your table. It start from the app. So just copy this one. Then you need to extract ID of your table because leads and property agents and developers have a different table ID. So you just need to select list that you need and copy this one and put instead of it. Okay, here we need to provide our token. So if you want to create uh your token just go to let's show you go to your builder hub then create a new token NA10 tutorial and we need to provide access for horizontal estate database and we need to scope like I'm proposed to add all of this record and base scheme. I think it's ready. So, let's create a token. Copy this token. Go to our NA10. Put this token. Okay, I think it's ready. Just click click on here. So, let's test it. So let's see we have output as you can see this is information about about our property we have a 100 record so it's like an array and we need to extract all of this record from this array because we have only one item currently. So what we need to do next, we just need to create a code and extract this. Okay, so we grab all of this data from our array. So currently we have 100 items. It means like we extract all of these items. Uh so the next step what we need to do is like a create a pine cone vector store from this data and use this pine vector store in in the future for our AI agent. Also I provided here a convert to file. So for example, if you want to export this one in the tesa and upload to our Google drive, but it's not a required step. Okay, let's create our pine core vector store. Add documents to vector store. So you need to have a pine core API account. So for example uh you just need to create your account in the system and then create an API key. Let's for example create a new one. is successfully. Uh then we need to use our index. Let's create from the scratch. Just let me show you. Go to database create index. We need to select text embed in three small. All of this information is okay. Uh let's take a name like property create index. Okay, our next step is to select this property from the list. Just wait. Maybe I need to refresh property. Okay, it's ready. So let's let's move to the next embedding. embed in open AI. We need to select the same three small that we selected before for our pine core. Move it here to the document. So default and we need to create our JSON file because this is a JSON we have a different fields and we need to extract our row and put into one string. Uh so what I'm proposed to do I'm proposed to create uh one sentence one long sentence that included all of this value and it will store inside the vector database. I have already prepared some code here. Just let me show you how it works. Okay. So I'm just for example I'm start to write in a sentence the property. So here I need a property name. So I'm just move my property name uh to this text field and I can see also how it looks uh in our database and you need just to create your sentence uh using this approach. So I just include all of these columns in one sentence. You can do for example in other way you can write a title for your table put the value and uh show it by comma. So if we have a lot of uh addresses like by comma it's also showed like this one. So in and our AI agent will read this one and make decision based on this data. Okay. Go to Z here. We also need to use a cursive character. Put here at 200. Okay, let's test our workflow. Okay, it's done. So, let's test how it looks in our index. So when we update our table we can see list of our addresses and property names. So for example, we can see the sentence that we created before like the property business bay gardens is currently available and we have this sentence for every record and it called a chunk like we have 100 chunks and this is vector will use in the future by AI agent to quickly identifi identify and search different property in database. So let's create an AI agent. So let's add a trigger. It would be a chat AI agent or let's use not an AI agent. Okay, sorry. So we have a chat uh okay the next one chat model it would be open AI memory we will use I don't know for example simple memory so we need like a context if you want to create a dialogue with a chat and we we'll need to chat remember our previous answers we need to create a context for example here we have a Five. Let's do like this one. Next one. It's a tool. So, uh let's use vector store questions answer tool. Create a property name. Then property vector store. We will use our pine cone vector store that we created before. So we just need to select our list with all of this property. Let's do this. Okay. Then we need to select our model open AI. We can select for O mini. For embedding we use embedding open AI also ambin small. Um let's check how it looks. I I think it looks okay. So we can start our chat for example open chat and ask what is the most expensive uh property in the list. So we start to ah I'm sorry we need to add a system message this instruction like hey please check my vector store to find an answer. So just copy this text. Let's do this again. We just provide instruction to check the vector store database. Okay. So we have the most expensive property in the listing ranches villa price 4 million. Let's check it. Okay. Yeah. Yeah. This is a correct. So this is like a approach how you can build vector database.