Yes please via the API…be great…so we understand how to use the API config inside botpress as the chat frontend….open to other chat bot frontend but botpress seems quite well supported/active…bonus points if you can give it conversational memory! 👍😉…ps there’s botpress cloud and self hosted…
You're welcome! Working on a discord where I will post much more thorough versions just in case I omitted things too much here, the discord version should have the answers. Also will add simple videos on how to deploy to different environments like railway etc.
🎯 Key Takeaways for quick navigation: 00:00 Creating an AI chatbot called Flowise that processes information from a provided website and uses the GPT-3.5 model. 00:26 Flowise allows fine-tuning the chatbot to the needs of a specific business or task, unlike pre-baked products with limited customization. 01:19 The chatbot sends queries to OpenAI GPT-3.5 by converting questions into vectors and accessing information from a vector database. 02:17 The chatbot's interaction with the vector database enables it to respond to queries about a website or PDF. 03:34 Flowise allows customization of the chatbot's appearance and functionality using JavaScript or Python code. 09:23 Flowise requires a server to run the application and can be deployed on platforms like Render. 10:15 The chatbot can be configured to retrieve information from websites, PDFs, or other sources using web scraping modules. 12:56 Templates in Flowise can be used to combine various APIs and create conversational retrieval models. 15:02 The chatbot can be easily embedded into websites using Webflow or other web development tools. 16:43 Flowise chatbot can be integrated with WordPress websites using web page builders like Elementor. 18:00 The video showcases a working example of the chatbot and encourages viewers to explore AI integration opportunities for their businesses. Made with HARPA AI
I'm using a PDF File node instead of the cheerio web scraper. My answers don't come out like yours do. I'm thinking it's because I'm using the PDF node and that I'm formatting it wrong, do you know a good way to formar the PDF to get better answers? Or is the problem the PDF node? Thanks, this video was really valuable!
I think it does help to just to test things out. One of my deployments is using pdf loader and it works okay. If you got to discord I have the area where people share what they are building and either me or other members will help you. If you take the screenshot that would be helpful
I think yes, you can even replace it with other icon if you want. Since you are the one who is hosting flowise you can make it look like anything you want, just requires a bit of JavaScript
So recently Render increased prices so I suggest to go with Railway instead, it is cheaper and it takes care of the .env file management from inside. You can change things using variables tab. Here is 1 click install for Flowise on Railway tinyurl.com/Flowise-Railway
@@aisimp Thank you for what you're doing. You've been very helpful. In render, I was having a weird issue when trying to create the web service. I had to add "yarn; yarn install; yarn build" in the Build Command section. Not sure if anyone else is having this issue but thought I would add it here just in case.
I don’t think so. You can still use the same deployment, for instance you deploy on railway and pay $5 to keep it running. Within this deployment you can have multiple flows and each can server a different website.
You need to use disk function in render , because free instance will pause after use and if you don’t have disk persistence func enabled it will reset every time. Here is the documentation docs.flowiseai.com/deployment/render
Thanks for the tutorial! Could you explain the Document loader called "Folder with Files" and how you set it up? I want to deploy a lot of PDF files in a folder and upload them but I don't understand what the URL / Path section means in the module. Could you please help?
Document loader is the langchain module that flowise abstracts. I will make a bubble video next that will go over the mechanics of file loading langchain does. They do abstract a lot of what happens under the hood so might be a bit confusing. I learned how to do it in Python first and then these tools started appearing so it is a bit easier for me to understand them. But yeah ,🔔 for the next video where I will make sure to go over the mechanics of the folder loader
Hi Simp- Absolutely love you’re channel, seriously! I’m so engaged! So much so that I a small question. Do I NEED a server to use flowise or can I just plug in the website and get on with it? I’m just asking because you mentioned that others can access your flowise. Please let me know. That part is a little intimidating for me. I’m very new to all of this. Thank you in advance!!! Please keep up the good work🤩😎👏🏾👏🏾 Oh and which do you prefer Langchain/flowis or chatbase? 😅
If you have little experience with anything technical and need the chatbot tomorrow. Get a chatbase for a month or two learn how to use Flowise ( so it is just software, it doesn’t work unless it is running on the computer that is connected to internet aka server. Render cost $7/ month) and then just stop paying for chatbase and use your flowise deployment.
I thought every time you chat with an upsert format it does run the whole process again. I've seen a second flowise that then just dips into the Vec DB after the fact. So one flow for upsert one flow for production retrieval? @@aisimp
@@aisimp if I want to build a chatbot for my website which will answer all queries of customers related to my product, do I need a memory for this? If yes, then which type of memory would you recommend?
I didn't understand how Langchain played a role in this. Also didn't really understand why u connected the web scraper. Is the goal that it collects all the website info and stores it in the vector database to 'teach' the bot what kind of services, posts, products etc. u have? Very cool video tho looking forward to trying it myself
So Flowise is just a GUI wrapper for Langchain. All of the functionality we have in Flowise is actually powered by Langchain. And regarding the webscraper exactly correct 👍, it is just a jargon that is common online to call collecting information from a website as “scrapping “.
The file is already in the GitHub, all you need is just change the .env.example to .env and change password or just use my railway template in the description it does it automatically on a railway instead of render
@@aisimp Thank you! But I can't make the flowise work... I did everything like you and when I talk to the chat it says: TypeError: Cannot read properties of undefined (reading 'replace') Thanks in advance!!
I have tried it but it scrapped data from the homepage only and didn't scrap data from the rest of the pages... But I want it collect information from the whole website... Can u please suggest me a solution... I ll really appreciate it
What we can do is create upsert module using bubble free tier and connect it flowise. I think it should provide options on how deep to crawl etc. What do you think?
@@RedPeaUnicorn I made a video on this topic. Check it out Build AI Server and AI-powered App for Personal or Business use. Flowise, Langchain, Flutter th-cam.com/video/4Am5JsEe6zo/w-d-xo.html
Hello Simp, The great video! I am concerned about exposing such details like chatflowid: when you embed them into HTML file after deployment. People can see this and copy-paste it to use somewhere else. Is there a way to hide it? Or Do you think it is okay to be exposed?
Good point, I thought about that, but given that you usually have your ai bot use respond only to data in the knowledgebase, this becomes not very useful to whomever will get the chatflow. But that's why I think having a layer in the middle like voiceflow or botpress provides that additional layer of security.
@@aisimp That is true, as long as I apply additional Params to my chatbot like”you are an agent of (this website )”it doesn’t work as good chatbot other websites. I make it unique as much as I can then. Thank you!
thats fairly easy, just need to fork the actuatl flowise bot, change it what you want it to look like and then when you are deployingt to your website change the repo to yours
Can you elaborate on how you see the integration to work. Give me a use case in few sentences and let’s see if I can come up with a decent way to integrate them
i was thinking of making it use as alternative to botpress manychat like chatbot scenario correct me if I am thinking in right way and if it can be done. if it can be done we can integrate it with Instagram and engage followers
Another fine video👍
Thank you. Quick question, what have you heard about botpress? Would you be interested in a video combining Botpress and Flowise together?
Yes please via the API…be great…so we understand how to use the API config inside botpress as the chat frontend….open to other chat bot frontend but botpress seems quite well supported/active…bonus points if you can give it conversational memory! 👍😉…ps there’s botpress cloud and self hosted…
Hello, thanks for the video. I have a question what if you want to get information from multiple websites (urls)
Thanks for the tutorial, really helpful
You're welcome! Working on a discord where I will post much more thorough versions just in case I omitted things too much here, the discord version should have the answers. Also will add simple videos on how to deploy to different environments like railway etc.
🎯 Key Takeaways for quick navigation:
00:00 Creating an AI chatbot called Flowise that processes information from a provided website and uses the GPT-3.5 model.
00:26 Flowise allows fine-tuning the chatbot to the needs of a specific business or task, unlike pre-baked products with limited customization.
01:19 The chatbot sends queries to OpenAI GPT-3.5 by converting questions into vectors and accessing information from a vector database.
02:17 The chatbot's interaction with the vector database enables it to respond to queries about a website or PDF.
03:34 Flowise allows customization of the chatbot's appearance and functionality using JavaScript or Python code.
09:23 Flowise requires a server to run the application and can be deployed on platforms like Render.
10:15 The chatbot can be configured to retrieve information from websites, PDFs, or other sources using web scraping modules.
12:56 Templates in Flowise can be used to combine various APIs and create conversational retrieval models.
15:02 The chatbot can be easily embedded into websites using Webflow or other web development tools.
16:43 Flowise chatbot can be integrated with WordPress websites using web page builders like Elementor.
18:00 The video showcases a working example of the chatbot and encourages viewers to explore AI integration opportunities for their businesses.
Made with HARPA AI
Ty! Really great stuff pal!
I'm trying to nail down the progressive learning from the chat history.
Which clip is that in?
Not sure what are you referring to. If it helps There is a full unedited version in the discord that you can look through.
Let me know
I'm using a PDF File node instead of the cheerio web scraper. My answers don't come out like yours do. I'm thinking it's because I'm using the PDF node and that I'm formatting it wrong, do you know a good way to formar the PDF to get better answers? Or is the problem the PDF node? Thanks, this video was really valuable!
I think it does help to just to test things out. One of my deployments is using pdf loader and it works okay. If you got to discord I have the area where people share what they are building and either me or other members will help you. If you take the screenshot that would be helpful
Wow ❤🎉 great video!
Thank you! 🤗
Awesome video. I have a question. can we place the flowise chatbot icon at the centre of the webpage or in any other position ?
I think yes, you can even replace it with other icon if you want. Since you are the one who is hosting flowise you can make it look like anything you want, just requires a bit of JavaScript
Not sure if i missed it but how do I create the .env file?
So recently Render increased prices so I suggest to go with Railway instead, it is cheaper and it takes care of the .env file management from inside. You can change things using variables tab. Here is 1 click install for Flowise on Railway tinyurl.com/Flowise-Railway
Timestamps for this video @aisimp
0:00 - Introduction
0:44 - Stack Overview
3:15 - Flowise Introduction
5:15 - Render Introduction
5:41 - Forking Flowise
5:58 - Connecting Render To Your Flowise Fork
6:50 - The Importance Of Creating a .ENV file
9:18 - Deploying Flowise Environment
9:45 - Flowise Marketplace "Conversational Retrieval Q&A Chain" Template
10:08 - Understanding The "Conversational Retrieval Q&A Chain" Nodes
11:55 - Setting Up/Configuring "Conversational Retrieval Q&A Chain"
12:02 - Recursive Character Text Splitter
12:13 - Remove .txt and replace it with Cheerio Web Scrapper
13:09 - OpenAI (Chat OpenAI and OpenAI embeddings)
13:34 - Pincone
13:59 - Conversational Retrieval Q&A Chain "Additional Parameters
15:58 - Embedding Flowise Chatbot Into Your Website
16:12 - Embedding Into Webflow Website
16:26 - Testing Webflow Integration
16:54 - Embedding Into Worpress Website
17:47 - Outro (Like & Subscribe)
Hey thanks for adding the timestamps 🙏🏻
@@aisimp Thank you for what you're doing. You've been very helpful.
In render, I was having a weird issue when trying to create the web service. I had to add "yarn; yarn install; yarn build" in the Build Command section. Not sure if anyone else is having this issue but thought I would add it here just in case.
Hello @aisimp
How will this work if I have multiple and different websites ?
Will that require a new box with web scrapper ?
I don’t think so. You can still use the same deployment, for instance you deploy on railway and pay $5 to keep it running. Within this deployment you can have multiple flows and each can server a different website.
I can’t find my saved chatflow when i reopen my flowise web service on render. Is this expected when using the free plan?
You need to use disk function in render , because free instance will pause after use and if you don’t have disk persistence func enabled it will reset every time. Here is the documentation docs.flowiseai.com/deployment/render
Thank you for the quick response! Much appreciated.
Very Cool and amazing video, I have a question can we have another option to use instead of Pinecone Upsert?
So Pinecone upsert is because I had it already. But you can use Supabase, z*something, m*something and other vector databases
@@aisimp Thank you, Sir. I am waiting for the next video.
Thanks for the tutorial! Could you explain the Document loader called "Folder with Files" and how you set it up? I want to deploy a lot of PDF files in a folder and upload them but I don't understand what the URL / Path section means in the module. Could you please help?
Document loader is the langchain module that flowise abstracts. I will make a bubble video next that will go over the mechanics of file loading langchain does. They do abstract a lot of what happens under the hood so might be a bit confusing. I learned how to do it in Python first and then these tools started appearing so it is a bit easier for me to understand them. But yeah ,🔔 for the next video where I will make sure to go over the mechanics of the folder loader
@@aisimp Thank you!
Hi Simp- Absolutely love you’re channel, seriously! I’m so engaged! So much so that I a small question. Do I NEED a server to use flowise or can I just plug in the website and get on with it? I’m just asking because you mentioned that others can access your flowise. Please let me know. That part is a little intimidating for me. I’m very new to all of this. Thank you in advance!!! Please keep up the good work🤩😎👏🏾👏🏾
Oh and which do you prefer Langchain/flowis or chatbase? 😅
If you have little experience with anything technical and need the chatbot tomorrow. Get a chatbase for a month or two learn how to use Flowise ( so it is just software, it doesn’t work unless it is running on the computer that is connected to internet aka server. Render cost $7/ month) and then just stop paying for chatbase and use your flowise deployment.
Does this not upsert the document every time you save the chat flow
To be honest number of vectors stayed the same between saves so probably not
I thought every time you chat with an upsert format it does run the whole process again. I've seen a second flowise that then just dips into the Vec DB after the fact. So one flow for upsert one flow for production retrieval?
@@aisimp
Can i use AWS and Azure instead of render ?
Of course
@@aisimpThank you for the reply,
One more question, is there an option to add memory to this chatbot?? because I see memory is not used in this video
@@farhanafridi8694 yeah there are many different types of memory modules for Flowise, each has its own use cases
@@aisimp if I want to build a chatbot for my website which will answer all queries of customers related to my product, do I need a memory for this? If yes, then which type of memory would you recommend?
I didn't understand how Langchain played a role in this. Also didn't really understand why u connected the web scraper. Is the goal that it collects all the website info and stores it in the vector database to 'teach' the bot what kind of services, posts, products etc. u have? Very cool video tho looking forward to trying it myself
So Flowise is just a GUI wrapper for Langchain. All of the functionality we have in Flowise is actually powered by Langchain. And regarding the webscraper exactly correct 👍, it is just a jargon that is common online to call collecting information from a website as “scrapping “.
It is fun to play with Bubble and Flowise together, because you can get 80% of functionality from zero coding approach
@@aisimp Thank you!!!
Amazing tutorial! You just didn´t show how to upload the .env file :(
The file is already in the GitHub, all you need is just change the .env.example to .env and change password or just use my railway template in the description it does it automatically on a railway instead of render
@@aisimp Thank you! But I can't make the flowise work... I did everything like you and when I talk to the chat it says:
TypeError: Cannot read properties of undefined (reading 'replace')
Thanks in advance!!
I have tried it but it scrapped data from the homepage only and didn't scrap data from the rest of the pages... But I want it collect information from the whole website... Can u please suggest me a solution... I ll really appreciate it
What we can do is create upsert module using bubble free tier and connect it flowise. I think it should provide options on how deep to crawl etc. What do you think?
Hey-I’m having the same issue. 😕
@@RedPeaUnicorn I made a video on this topic. Check it out Build AI Server and AI-powered App for Personal or Business use. Flowise, Langchain, Flutter
th-cam.com/video/4Am5JsEe6zo/w-d-xo.html
Hello Simp, The great video! I am concerned about exposing such details like chatflowid: when you embed them into HTML file after deployment. People can see this and copy-paste it to use somewhere else. Is there a way to hide it? Or Do you think it is okay to be exposed?
Good point, I thought about that, but given that you usually have your ai bot use respond only to data in the knowledgebase, this becomes not very useful to whomever will get the chatflow. But that's why I think having a layer in the middle like voiceflow or botpress provides that additional layer of security.
@@aisimp That is true, as long as I apply additional Params to my chatbot like”you are an agent of (this website )”it doesn’t work as good chatbot other websites.
I make it unique as much as I can then. Thank you!
how do i remove the flowise link at the bottom of the chatbox?
thats fairly easy, just need to fork the actuatl flowise bot, change it what you want it to look like and then when you are deployingt to your website change the repo to yours
How can do it with next js 13?
Working on this right now, using nextjs app structure , not very familiar with it so taking a bit extra time 🥵
r u a frenchie?
Studied french for 3yrs in high school, so yeah CENT pourcent 🇫🇷
Exactly what I was looking for! Thanks! 🙏🏼 Can I connect with you on FB?
Hey I don’t really use FB, but feel free to connect on Twitter @aisimpdaily
how can we integrated with whatsapp instagram or WhatsApp telegram
Can you elaborate on how you see the integration to work. Give me a use case in few sentences and let’s see if I can come up with a decent way to integrate them
i was thinking of making it use as alternative to botpress manychat like chatbot scenario correct me if I am thinking in right way and if it can be done. if it can be done we can integrate it with Instagram and engage followers