Automated PDF Analysis: Using ChatGPT & Zapier For Any Industry | Tutorial

Поділитися
Вставка
  • Опубліковано 15 гру 2024

КОМЕНТАРІ • 55

  • @Corbin_Brown
    @Corbin_Brown  Рік тому +10

    Very important, everyone! Make sure you save the new file from Google Docs in a new folder. This will ensure that you don't create a loop where it gets triggered again when the document is formed and goes through the same process.

    • @wookawsh
      @wookawsh 7 місяців тому

      Very helpful content. To clarify, to make this process repeat for every new file that is created it needs to go into its own new folder as well. So do we need an extra step to create a new folder >> File goes into that new folder? Thanks so much.

  • @humphuk
    @humphuk Рік тому +2

    Corbin - These are excellent tutorials. I came here with a specific user case that I thought of today. You have answered it directly - and given me inspiration for so much more. Once I have digested and tried - I will be back for more. Thank you

  • @Corbin_Brown
    @Corbin_Brown  Рік тому

    Navigate to key moments👇
    made via tubestamp.com
    01:32 - Demonstrating automation process with a legal PDF example.
    09:20 - Testing code and showing output of chunking a legal document.
    13:50 - Showing compressed output of a legal contract using GPT-
    15:41 - Explanation on getting a one-sentence summary using GPT-
    21:05 - Creating sections and summaries for larger documents.
    24:17 - Live demonstration of the process with successful results.
    26:11 - Recap of the automated process for compressing and finding data within PDFs.
    Recap by TubeStamp ✏️

  • @Corbin_Brown
    @Corbin_Brown  Рік тому +6

    Here's the code we used in today's tutorial:
    Word Output:
    const response = await fetch(inputData["text"]);
    const text = await response.text();
    const words = text.split(" ");
    output = { result: words.slice(0, 20000).join(" ") };
    Chunk Maker:
    const wordLimit = 5000;
    const words = inputData.text.split(/\s+/); // Split by whitespace
    const chunks = [];
    let currentChunk = [];
    for (let word of words) {
    if (currentChunk.length + 1 > wordLimit) {
    chunks.push(currentChunk.join(' '));
    currentChunk = [];
    }
    currentChunk.push(word);
    }
    if (currentChunk.length > 0) {
    chunks.push(currentChunk.join(' '));
    }
    // Prepend unique identifiers
    const identifiedChunks = chunks.map((chunk, index) => `ID${index + 1}_${chunk}`);
    // Return the chunks with identifiers
    output = {
    chunk1: identifiedChunks[0] || "",
    chunk2: identifiedChunks[1] || "",
    chunk3: identifiedChunks[2] || ""
    };

    • @roberthudson9342
      @roberthudson9342 Рік тому

      I keep getting a Failed to run JavaScript code error. Can you advise on this?

    • @Corbin_Brown
      @Corbin_Brown  Рік тому

      @@roberthudson9342 Go ahead and reply to this comment with the exact error so I can provide better insight. Typically, errors could be associated with how you named your input data. Make sure it's called 'text,' and the data point is 'File txt' (Exists but is not shown)

    • @Omar-AI
      @Omar-AI Рік тому +1

      @@Corbin_Brown Hi, I got the same error "Failed to run your JavaScript code
      2023-11-03T17:26:31.834Z 8cc47817-727d-4f01-b110-9e7982e09ea4 Task timed out after 1.00 seconds"
      Even If i use same inputs as you did in the video (Input Data = Text and 2. File Txt: Exits but not shown.
      Can you advice on this please ?
      Thanks a lot !

    • @Corbin_Brown
      @Corbin_Brown  Рік тому

      ​@@Omar-AI Hey! Make sure that for the input data, you spell it like this 'text'. Capitalization is important in the context of code, as it won't be able to read the data point if you put 'Text'. Hope this helps!

    • @Omar-AI
      @Omar-AI Рік тому

      @@Corbin_Brown I just double checked the spelling and It seems good. I'm really stuck because of this error :(

  • @nerdsdoitbettervid
    @nerdsdoitbettervid 5 місяців тому +3

    Where is the javaScript?

  • @MyXRLearning
    @MyXRLearning 11 місяців тому +1

    Thank you so much. Your videos are so helpful!

  • @TheMrsSusso
    @TheMrsSusso 7 місяців тому

    Thank you very much. You just made my business much more efficient. God bless you and your journey!

  • @TheHealmobile
    @TheHealmobile 3 місяці тому +2

    Great video!

  • @ziadnahdi4343
    @ziadnahdi4343 Рік тому +1

    شكرا جزيلا هذا كلام من احسن الفيديوهات التي رايتها شكرا وتابع العمل

  • @EricHardek
    @EricHardek Рік тому

    This is very cool I like how you used 3.5 16k to compress and the GPT 4 to compress again. This is great content!

  • @joshfire8812
    @joshfire8812 2 місяці тому +1

    This is so cool!

  • @Corbin_Brown
    @Corbin_Brown  Рік тому +1

    Hi all! Exciting news - we've integrated a PDF Reader right into ChatGPT. It's built to handle large PDFs with ease and is completely free for you to use.
    👉 GPT [PDF Reader] - chat.openai.com/g/g-QSh6KHL3S-pdf-reader

    • @drew_scottsdale
      @drew_scottsdale 3 місяці тому +1

      are we able to use this with over zapier

  • @SKTeuber
    @SKTeuber 2 місяці тому

    I simply can not take my results from the Chat GPT Conversation and put it into a Google Sheet or Excel. Keeps saying invalid image.

  • @hammadyounas2688
    @hammadyounas2688 5 місяців тому

    I have the pdf document. i dnt have the text extract from it. How can i do that?

  • @rehatvirk1807
    @rehatvirk1807 9 місяців тому

    im only getting the url instead of the text in the code result

  • @ivanazima8118
    @ivanazima8118 8 місяців тому

    Hi Corbin, I have tried to repeat the steps with One Drive but I can not get the data out of the PDF. Can you please help me or make another video about One Drive integrations?

  • @thestudentshuttledmv5090
    @thestudentshuttledmv5090 13 днів тому

    Looks like the zapier code is giving this error for individuals running it: Failed to run your code
    Your code had an error. See details below:
    Error: Error: only absolute urls are supported
    at (line 1, column 28)
    Anyone having this same issue?

  • @learningstuff5679
    @learningstuff5679 Рік тому +1

    With all of the "Chat With PDF" ai platforms that are currently available, would you say that this is the safest way for sensitive documents like legal, finance etc?

    • @Corbin_Brown
      @Corbin_Brown  Рік тому +1

      Yes, when you use the ChatGPT front end, your data is not secure and can be accessed by either the Plugin or OpenAI. Therefore, using the OpenAI API is more secure, as they have stated that they do not have access to the prompts being used in this context.

  • @WolfTheatres362
    @WolfTheatres362 6 місяців тому

    Thank you for posting the video. Is it possible to create a ZAP that would allow the Chatbot PDF reader you created to automatically read a PDF after the initial trigger of say uploading the pdf is google drive?

  • @AngieEspinal-c6r
    @AngieEspinal-c6r Рік тому

    The java script gives me an Error. Help.

  • @learningstuff5679
    @learningstuff5679 Рік тому

    Corbin, can you make a video on setting up and onboarding an automation client? getting passwords, creating accounts, the process of onboarding them, reporting etc? thank you

  • @masdeia
    @masdeia 10 місяців тому +1

    Hello Corbin, do you have any video, or any suggest me about how to create a gpt that has the capability to review, it can be either 100 pdfs documents or more, and display important details of those pdfs files, those pdfs will be in google cloud, but will keep adding more and more, the output can be set in the same gpt, in a table or simple text no problem.

    • @Corbin_Brown
      @Corbin_Brown  10 місяців тому

      As of now, I do not have a video like the one you described, but I could consider making one in the future!

    • @masdeia
      @masdeia 10 місяців тому

      @@Corbin_Brown thanks, do You know if it is possible to pass a folder containing múltiple PDF files, instead of just one PDF in askyourpdf or any other plugin

  • @grant-taleck
    @grant-taleck 11 місяців тому +3

    You mind copying and pasting the javascript here? I couldn't find it in the comments?

    • @Corbin_Brown
      @Corbin_Brown  11 місяців тому

      Check out this alternative way to get text from a file now!
      ua-cam.com/video/D8-6HELz6wA/v-deo.html&ab_channel=Corbinai

    • @grant-taleck
      @grant-taleck 11 місяців тому

      actually, I figure the better way to do this now is through the method you mention in this video? ua-cam.com/video/bnCDtIYCTsA/v-deo.html

  • @joshuabowman442
    @joshuabowman442 Рік тому +1

    We receive password-protected PDFs. Our client does not provide the masterPDF password, only the password to open the document. Corbin, do you know if it is possible to add a step in the zap that will fill in the password (not necessarily remove the password, just fill it in so we can open the PDF and convert it to a google doc so we can read the data as you show in the tutorial?)

    • @Corbin_Brown
      @Corbin_Brown  Рік тому +1

      In this context, it may be possible, and it would all depend on whether the PDF software gives us access to their API documentation. If so, in theory, one could input a password and gain access to the data.

    • @Fadayko23
      @Fadayko23 8 місяців тому +1

      You can use a tool like Bardeen to create web scrape template for the content of the PDF via a web URL or a right-click action..from there sky is the limit

  • @jacksandra1
    @jacksandra1 Рік тому

    Hi Corbin, I am getting an error with the first part of the javascript "Only absolute URLs are supported" can you help me resolve this error please?

    • @Corbin_Brown
      @Corbin_Brown  Рік тому

      This is typically associated with the formatting of the URL. Make sure to use the entire security measure of

  • @frederikblegvadandersen5151

    Hey Corbin - thank you very much. Spoton 🙏

  • @learningstuff5679
    @learningstuff5679 Рік тому

    Also would love to start seeing timestamps in your videos.

  • @stanTrX
    @stanTrX 2 місяці тому

    Whats Zapier, never heard that project before. İs it still active?

  • @marcink79
    @marcink79 Рік тому

    In 3. step
    Failed to run your JavaScript code
    2023-11-07T21:25:10.320Z ea21f49d-ebf9-45de-bc6c-a57322b0de4c Task timed out after 1.01 seconds

    • @Corbin_Brown
      @Corbin_Brown  Рік тому

      Usually correlated with misnaming of variables, if the issue persists, try to output lower amounts of text to see if everything else is correct.

    • @mariodemartino1097
      @mariodemartino1097 Рік тому

      I faced the same issue as @marcink79 I typed this code const response = await fetch(inputData["text"]);
      const text = await response.text();
      const words = text.split(" ");
      output = { result: words.slice(0, 20000).join(" ") }; is it the correct one?

    • @jacksandra1
      @jacksandra1 Рік тому

      Once I upgraded to a paid subscription the code worked