Step-by-Step Tutorial | Build an AI Agent with n8n and Pinecone (NO CODE!!)

Поділитися
Вставка
  • Опубліковано 15 гру 2024

КОМЕНТАРІ • 61

  • @AI-GPTWorkshop
    @AI-GPTWorkshop  Місяць тому

    📚 Find all the Resources and Live daily Q&A calls in my Skool community 👉 www.skool.com/aiworkshop

  • @merakshitrading9216
    @merakshitrading9216 27 днів тому

    Best demo to see the collaboration between workflow, vector db and other applications

  • @sirishkumar-m5z
    @sirishkumar-m5z 3 місяці тому +1

    For anyone wishing to develop an AI agent without delving into coding, this guide is immensely helpful. Ideal for novices! #NoCode #AIAgent

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      Thanks. Glad you found it helpful. Much more to come :)

    • @hajduadam
      @hajduadam Місяць тому

      ​@@AI-GPTWorkshop
      can you add a video how to use existing postgres sql as a data source to integrate with?

  • @AI-GPTWorkshop
    @AI-GPTWorkshop  3 місяці тому +6

    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

  • @simfor
    @simfor 18 днів тому

    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

  • @lalpremi
    @lalpremi 14 годин тому

    Thank you for sharing, have a great day 🙂

  • @SoloJetMan
    @SoloJetMan Місяць тому

    good stuff dude - all of this is very good timing for my project!

  • @calebweintraub1
    @calebweintraub1 3 місяці тому +1

    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.

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      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.

  • @NmkDozer
    @NmkDozer 3 місяці тому

    This video is a gold mine, keep going like this

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      Thanks. Glad you enjoyed it. A lot more to come.

  • @BlueBirdAIstudio
    @BlueBirdAIstudio 2 місяці тому +3

    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?

  • @MarcIbizaMena
    @MarcIbizaMena 5 днів тому

    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?

  • @FutureOfCinema
    @FutureOfCinema Місяць тому

    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

  • @AyS-kj4ok
    @AyS-kj4ok 5 днів тому

    Thank you. Does the embedding run each time we question the AI? What if we need to vectorize and embedd a document only once?

  • @whatsgoingonwithyp275
    @whatsgoingonwithyp275 2 місяці тому

    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.

  • @careersunlimitergv
    @careersunlimitergv Місяць тому

    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.

  • @jeelanshahtlyr6076
    @jeelanshahtlyr6076 3 місяці тому +1

    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.

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  2 місяці тому

      hmm those are great questions. I will test out a few things and let you know if find the answer

  • @MAHMOODALABBAS
    @MAHMOODALABBAS 3 місяці тому +1

    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?

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      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

  • @masyitahmustapha
    @masyitahmustapha 6 днів тому

    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....

  • @hajduadam
    @hajduadam Місяць тому

    @AI-GPTWorkshop can you add a video how to use existing postgres sql as a data source to integrate with?

  • @potenciaractivo
    @potenciaractivo 3 місяці тому +1

    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! 🙏

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  2 місяці тому

      try GPT-4o mini it's a lot cheaper and yes you can upload google sheets

  • @agsvk-com
    @agsvk-com 2 місяці тому

    Does pinecone get expensive compared to something like chroma?

  • @JoaoCarlos-bp8ip
    @JoaoCarlos-bp8ip 2 місяці тому

    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"

  • @WilfredoEliasPimentelSerrano
    @WilfredoEliasPimentelSerrano 3 місяці тому

    Muy interesante, felicitaciones

  • @InDeathUnchainedTopTips
    @InDeathUnchainedTopTips 3 місяці тому +2

    Can you do a video on just how to embed the flow on a website?

  • @seanolivas9148
    @seanolivas9148 3 місяці тому

    I hear supabase is great because you can use it for both memory and database. What's your thoughts?

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  2 місяці тому

      yah I like supabase too. it's a great alternative since it was build from ground up for AI

  • @maxtkacz3748
    @maxtkacz3748 3 місяці тому

    Super insightful, nice!

  • @macaronidream
    @macaronidream 2 місяці тому

    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

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  2 місяці тому +1

      yup planning on it. stay tuned

    • @Vbastin-p2r
      @Vbastin-p2r 2 місяці тому

      And your website is running on ??

  • @ifnotnowwhen2811
    @ifnotnowwhen2811 3 місяці тому

    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?

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  2 місяці тому

      try increasing the chunk size for bigger documents and also make sure you use a good embedding model

  • @passiveecomhq
    @passiveecomhq 3 місяці тому

    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
      @AI-GPTWorkshop  2 місяці тому

      Thanks. are you asking if you can upload a folder to pinecone or just multiple files at once?

    • @tmcgann
      @tmcgann 2 місяці тому +2

      @@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?

    • @passiveecomhq
      @passiveecomhq 2 місяці тому

      Thanks for your response. I am asking how can upload multiple files at once.

  • @LeandroSartori2
    @LeandroSartori2 3 місяці тому +1

    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

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      ah damn you are right. I will put a comment and pin it so people are aware. Thanks for pointing that out.

  • @Lilia_AromaCoach
    @Lilia_AromaCoach 2 місяці тому

    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 ?

  • @Roxester
    @Roxester 3 місяці тому

    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.

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому

      yes you should be able to. you can utilize the filter node on n8n to separate the fields

    • @Roxester
      @Roxester 3 місяці тому

      @@AI-GPTWorkshop but using ai agent right?

  • @RamonTomzer
    @RamonTomzer 2 місяці тому

    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?

  • @BirdManPhil
    @BirdManPhil 2 місяці тому

    i keep getting /n all in my outputs

  • @notnyldier7150
    @notnyldier7150 4 дні тому

    so this is your own notebooklm kinda thing

  • @tirtir1401
    @tirtir1401 3 місяці тому

    please explain why pinecone service was chosen without any comments about the existing local in-memory?

    • @AI-GPTWorkshop
      @AI-GPTWorkshop  3 місяці тому +2

      Please watch the whole video and that question will be answered :)