LangChain Email Assistant (Developer Version)

Поделиться
HTML-код
  • Опубликовано: 27 янв 2025

Комментарии • 15

  • @sprobertson
    @sprobertson 13 дней назад +4

    Loving the developer focused video concept

  • @yazin-s3l
    @yazin-s3l 13 дней назад

    very helpful Harrison, thank you so much for walking us through the code!

  • @FiddlingwithAI
    @FiddlingwithAI 13 дней назад +2

    That's actually very helpful! Continue what you do

  • @cheahanqi964
    @cheahanqi964 9 дней назад

    TQ .... loving this video. Please more video like this. on developer focused.

  • @ozuromo
    @ozuromo 13 дней назад

    Thanks for the video!

  • @ocramas
    @ocramas 10 дней назад

    Not really sure how langchain works so, I got a question. How does this keep the email information still confidential? if it is calling openai. Langchain will run a RAG right? then build some "memory" over time, but how is the confidentiality handled?

    • @pcuciu
      @pcuciu 8 дней назад

      you'd have to use only locally deployed models to keep your emails private, and no langsmith api key, also don't deploy to langgraph platform; with the current design (what was shown in the video) open ai, anthropic, and langchain, they all see all your private emails

  • @mohamedkarim-p7j
    @mohamedkarim-p7j 13 дней назад +1

    Thank for sharing

  • @eddieotudor9432
    @eddieotudor9432 13 дней назад +2

    How sure are we that in the coming days we won't have another framework that's more "stable and controllable" than LangGraph? 😕🤔

  • @aifarmerokay
    @aifarmerokay 12 дней назад

    Can you cover multi knowledge based agent system where we are using multiple RAG ,example aws bedrock ,share point ,lancedb , web crawler,graph rag
    How to execute or which is suitable structure for this

  • @AndrewRevak
    @AndrewRevak 13 дней назад +2

    Didn't take long for google to flag your app as dangerous :)

  • @WinonaNagy
    @WinonaNagy 13 дней назад

    LangChain's AI Executive Assistant for emails seems innovative. Has anyone tried it and found it useful? Curious how effective the human-in-the-loop model is.

    • @sprobertson
      @sprobertson 13 дней назад

      human in the loop is just a pattern, but huge for anything where you don't want to trust a model to do something without authorization

    • @Rusputin-eo6tv
      @Rusputin-eo6tv 12 дней назад

      idk, you should try it out or wait a little bit for other youtubers to clarify it