Note: At 5:08 please make sure you click on PUBLISH APP under Publishing status otherwise it won't allow you to sign in with your google account on n8n
Ein wirklich großartiges video, ganz herzlichen dank. Das erste Mal dass jemand wirklich den Ablauf so erklärt dass ich es verstehe und der Weg auf dem du es erklärst es wirklich klasse. Das andere Mädchen wäre noch gewesen wenn du erklärt hättest was der Unterschied ist zwischen dem small Modell dass du verwendet hast und dem großen Modell was auch noch angeboten wird
As a non-programmer, I find this helpful. Excellent work. I'm interested in seeing measurable and/or perceivable differences in results when using various information retrieval methods through these tools. If you're aware of existing comparative analyses, it would be great to see how they demonstrate the distinctions between results from keyword matching, semantic search, and RAG when querying various types and volumes of datasets. I'm particularly curious about accuracy, speed, and handling of nuanced queries. A comparative analysis might reveal whether certain approaches are notably superior for specific tasks, beyond just cost considerations... My understanding is that unless the user is a small business and has tons of docs and tons of money to do endless API calls, vector-based expenses may be prohibitive--- but even if the resources exists, I wonder if it is overkill for some common use cases.
Thanks for the video, well done. Q1: Is there a way so that each document added to the GoogleDrive would automatically be added into the Pinecone vector DB? Q2: Could I add several document types (csv, text, images) into the same Drive folder and Pinecone would still be able to store the vector embeddings correctly?
Very good tutorial, a question: what type of format is best to upload to the drive, and vectorize its content? For example, to make a FAQ with questions and answers?
Thank you for the video! Very helpful! You showed how to retrieve from one file. Is there a way to reference multiple files in the same folder? I want to be able to ask questions about one file, then another question to a different file.
Thank you so much for you education; however, how do I put these agents to practical use and share these agents with my students to use in my network or connect to an LMS. Your feedback is greatly appreciated.
Excellent step by step tutorial I must admit. First, Is this workflow an alternate to using a RAG tool for similar document visual analysis. Second question is that in your given workflow if we wanted to compare two different documents (eg. Contracts) to either identify what are the missing articles in oldocument B when compared to document A. Or to ask the agent to elaborate on any implications if a particular addition was made to document B when compared to document A. Such a use case is mostly used in the legal industry, hence how would we go about such a comparison.
This is great and very insightful. Is there a way to use multiple files simultaneously or set up a process to keep a file updated with the latest data? For example, could Airtable be used as a source instead of Google Drive, with Airtable cells updated daily and n8n handling the embedding to vector storage?
Thanks brother. That's a great question. Yes I believe it can. n8n has an integration with Airtable. You just have to add another node in n8n to upload the new data regurlary into pinecone to automate the whole process
this video is so great!! but i have one question regarding on the data. my documents/data are all in json files. i need to use those for letting the AI agent to read those JSON files and summarise what is in the json file. how do i do that?? im currently using webhook trigger as the first node but starting from there, i dont have any idea....
Hi! Can you please make a video to connect with sheets to retriev products information. And anotheer question: This only works well with GPT-4o ? Because that model is extremely expensive? thank you! 🙏
What version of n8n are you using? Mine is 1.60.1 and I can't see Vector Store Retriever as Tool, but Vector Store Tool and it has 2 connections, "Vector Store" and "Model"
can you please do a video on how to deploy a chatbot like the one you created to a website. if not can you direct me in the right direction to do this myself
Thanks a $MM for this. I have been fumbling trying to do similair. One issue and One question. Issue I tried using the NVIdia anual reprot PDF and it either breaks or never loads. Is there a way to adjust size limit or other. The question. I would love to be able to search an assigned folder / data dump, as opposed to 1 doc at a time. Any pointers on doing that?
Great video very helpful. So I subscribed I like your content. Worked just as you described. I want to download more than one file from my Google Drive. I tried putting the files in a folder and was unsure of the settings. Can you help?
@@AI-GPTWorkshop Thank you for the great video. I have the same question. I want to upload many files from Google Drive into Pinecone for retrieval but couldn't figure out how to upload an entire folder, only an individual file. Is there a way to do this rather than by uploading each file separately to Pinecone?
I think you forgot to mention that you need to publish the APP while you create the google cloud OAuth key, otherwise it doesn't allow to Sign with your google account on N8N, I got stuck there, It says the APP wasn't published and I was not part of the Test list
hey brother . was going with you step by step . and got a problem at google drive connection . 1. "from list"- it was not showing me anything . 2. there was a message could not connect . then I selected "by URL" - then I got another message unable to connect without access TOKEN. ... so im stuck . will go and try to connect to another google drive .. AY ADVICE FROM YOUR SIDE ?
Can I create an AI agent that connects to a Google Sheet containing 8 million random entries, including dentists, dermatologists, HVAC professionals, and more? Based on specific input, the AI agent should generate a separate sheet for each category. For example, if the input is 'dentist,' a sheet specifically for dentists should be created.
so I have managed to get the chatbot embedded on a website, but when it responds, it responds in json. { "response": { "text": "I don't know." } }. anyone else run into this?
📚 Find all the Resources and Live daily Q&A calls in my Skool community 👉 www.skool.com/aiworkshop
Best demo to see the collaboration between workflow, vector db and other applications
For anyone wishing to develop an AI agent without delving into coding, this guide is immensely helpful. Ideal for novices! #NoCode #AIAgent
Thanks. Glad you found it helpful. Much more to come :)
@@AI-GPTWorkshop
can you add a video how to use existing postgres sql as a data source to integrate with?
Note: At 5:08 please make sure you click on PUBLISH APP under Publishing status otherwise it won't allow you to sign in with your google account on n8n
Ein wirklich großartiges video, ganz herzlichen dank. Das erste Mal dass jemand wirklich den Ablauf so erklärt dass ich es verstehe und der Weg auf dem du es erklärst es wirklich klasse. Das andere Mädchen wäre noch gewesen wenn du erklärt hättest was der Unterschied ist zwischen dem small Modell dass du verwendet hast und dem großen Modell was auch noch angeboten wird
Thank you for sharing, have a great day 🙂
good stuff dude - all of this is very good timing for my project!
As a non-programmer, I find this helpful. Excellent work. I'm interested in seeing measurable and/or perceivable differences in results when using various information retrieval methods through these tools. If you're aware of existing comparative analyses, it would be great to see how they demonstrate the distinctions between results from keyword matching, semantic search, and RAG when querying various types and volumes of datasets. I'm particularly curious about accuracy, speed, and handling of nuanced queries. A comparative analysis might reveal whether certain approaches are notably superior for specific tasks, beyond just cost considerations... My understanding is that unless the user is a small business and has tons of docs and tons of money to do endless API calls, vector-based expenses may be prohibitive--- but even if the resources exists, I wonder if it is overkill for some common use cases.
yah that's a good point. To be honest I don't have that analysis but if you find out more info let me know. You got me curious.
This video is a gold mine, keep going like this
Thanks. Glad you enjoyed it. A lot more to come.
Thanks for the video, well done. Q1: Is there a way so that each document added to the GoogleDrive would automatically be added into the Pinecone vector DB? Q2: Could I add several document types (csv, text, images) into the same Drive folder and Pinecone would still be able to store the vector embeddings correctly?
Very good tutorial, a question: what type of format is best to upload to the drive, and vectorize its content? For example, to make a FAQ with questions and answers?
The answer is "I don't know" about every time that I don't make a question with the exact content of the PDFs :D
Thank you. Does the embedding run each time we question the AI? What if we need to vectorize and embedd a document only once?
Thank you for the video! Very helpful! You showed how to retrieve from one file. Is there a way to reference multiple files in the same folder? I want to be able to ask questions about one file, then another question to a different file.
Thank you so much for you education; however, how do I put these agents to practical use and share these agents with my students to use in my network or connect to an LMS. Your feedback is greatly appreciated.
Excellent step by step tutorial I must admit. First, Is this workflow an alternate to using a RAG tool for similar document visual analysis. Second question is that in your given workflow if we wanted to compare two different documents (eg. Contracts) to either identify what are the missing articles in oldocument B when compared to document A. Or to ask the agent to elaborate on any implications if a particular addition was made to document B when compared to document A. Such a use case is mostly used in the legal industry, hence how would we go about such a comparison.
hmm those are great questions. I will test out a few things and let you know if find the answer
This is great and very insightful. Is there a way to use multiple files simultaneously or set up a process to keep a file updated with the latest data? For example, could Airtable be used as a source instead of Google Drive, with Airtable cells updated daily and n8n handling the embedding to vector storage?
Thanks brother. That's a great question. Yes I believe it can. n8n has an integration with Airtable. You just have to add another node in n8n to upload the new data regurlary into pinecone to automate the whole process
this video is so great!! but i have one question regarding on the data. my documents/data are all in json files. i need to use those for letting the AI agent to read those JSON files and summarise what is in the json file. how do i do that?? im currently using webhook trigger as the first node but starting from there, i dont have any idea....
@AI-GPTWorkshop can you add a video how to use existing postgres sql as a data source to integrate with?
Hi! Can you please make a video to connect with sheets to retriev products information. And anotheer question: This only works well with GPT-4o ? Because that model is extremely expensive? thank you! 🙏
try GPT-4o mini it's a lot cheaper and yes you can upload google sheets
Does pinecone get expensive compared to something like chroma?
What version of n8n are you using? Mine is 1.60.1 and I can't see Vector Store Retriever as Tool, but Vector Store Tool and it has 2 connections, "Vector Store" and "Model"
Muy interesante, felicitaciones
Can you do a video on just how to embed the flow on a website?
You mean embedded the chat window to a website ?
@@AI-GPTWorkshop Yes
Agree🎉
I hear supabase is great because you can use it for both memory and database. What's your thoughts?
yah I like supabase too. it's a great alternative since it was build from ground up for AI
Super insightful, nice!
Thanks Max :)
can you please do a video on how to deploy a chatbot like the one you created to a website. if not can you direct me in the right direction to do this myself
yup planning on it. stay tuned
And your website is running on ??
Thanks a $MM for this. I have been fumbling trying to do similair. One issue and One question. Issue I tried using the NVIdia anual reprot PDF and it either breaks or never loads. Is there a way to adjust size limit or other. The question. I would love to be able to search an assigned folder / data dump, as opposed to 1 doc at a time. Any pointers on doing that?
try increasing the chunk size for bigger documents and also make sure you use a good embedding model
Great video very helpful. So I subscribed I like your content. Worked just as you described. I want to download more than one file from my Google Drive. I tried putting the files in a folder and was unsure of the settings. Can you help?
Thanks. are you asking if you can upload a folder to pinecone or just multiple files at once?
@@AI-GPTWorkshop Thank you for the great video. I have the same question. I want to upload many files from Google Drive into Pinecone for retrieval but couldn't figure out how to upload an entire folder, only an individual file. Is there a way to do this rather than by uploading each file separately to Pinecone?
Thanks for your response. I am asking how can upload multiple files at once.
I think you forgot to mention that you need to publish the APP while you create the google cloud OAuth key, otherwise it doesn't allow to Sign with your google account on N8N, I got stuck there, It says the APP wasn't published and I was not part of the Test list
ah damn you are right. I will put a comment and pin it so people are aware. Thanks for pointing that out.
hey brother . was going with you step by step . and got a problem at google drive connection . 1. "from list"- it was not showing me anything . 2. there was a message could not connect . then I selected "by URL" - then I got another message unable to connect without access TOKEN. ... so im stuck . will go and try to connect to another google drive .. AY ADVICE FROM YOUR SIDE ?
reconnect your google account again
Can I create an AI agent that connects to a Google Sheet containing 8 million random entries, including dentists, dermatologists, HVAC professionals, and more? Based on specific input, the AI agent should generate a separate sheet for each category. For example, if the input is 'dentist,' a sheet specifically for dentists should be created.
yes you should be able to. you can utilize the filter node on n8n to separate the fields
@@AI-GPTWorkshop but using ai agent right?
so I have managed to get the chatbot embedded on a website, but when it responds, it responds in json. { "response": { "text": "I don't know." } }. anyone else run into this?
hmm is your workflow activated?
i keep getting /n all in my outputs
so this is your own notebooklm kinda thing
please explain why pinecone service was chosen without any comments about the existing local in-memory?
Please watch the whole video and that question will be answered :)