What is Agentic AI? Important For GEN AI In 2025

00:22:35
https://www.youtube.com/watch?v=xOS0BhhdUbo

Resumo

TLDRIn this video, Krish Naak discusses the advancements in generative AI and introduces the concept of agentic AI. Agentic AI represents autonomous systems that aim to achieve specific objectives without human oversight, contrasting it with generative AI, which primarily focuses on content generation. The video covers various frameworks like Lang Chain, F Data, and Microsoft Autogen that facilitate the development of agentic AI applications. Multiple examples illustrate the functionality of agentic AI, particularly in finance, showcasing its ability to perform complex tasks and provide data-driven recommendations.

Conclusões

  • 🚀 Agentic AI is an evolution of generative AI, focusing on autonomous task execution.
  • 🤖 It allows for complex workflows without human intervention.
  • 🔧 Frameworks like F Data and Lang Chain are key tools for building agentic AI apps.
  • 📈 Agentic AI has significant applications in finance, addressing market queries.
  • 🎯 The goal of agentic AI is to deliver specific business outcomes.
  • 💡 Understanding the differences between generative and agentic AI is crucial for developers.

Linha do tempo

  • 00:00:00 - 00:05:00

    Introduction to the channel and a summary of 2024, focusing on advancements in generative AI and the rise of large language models (LLMs). Discussion about the emergence of Agentic AI and the intention to elaborate on its capabilities and frameworks in the video.

  • 00:05:00 - 00:10:00

    Explanation of the differences between generative AI and agentic AI; while generative AI focuses on content creation through user queries and prompts, Agentic AI systems operate autonomously, working toward specific goals without requiring constant human input. Emphasis on integrating multiple tools to enhance functionality.

  • 00:10:00 - 00:15:00

    Introduction to practical examples and illustrations of agentic AI applications, emphasizing the complex workflows that can be generated to accomplish tasks like financial decision-making. A clear distinction between how agentic AI can autonomously manage various agents compared to generative AI.

  • 00:15:00 - 00:22:35

    Insight into various frameworks available for developing agentic AI applications, such as F data, Microsoft Autogen, and Langflow, and a demonstration of creating an end-to-end agentic AI application that interacts with external sources like stock market APIs to provide financial recommendations.

Mostrar mais

Mapa mental

Vídeo de perguntas e respostas

  • What is agentic AI?

    Agentic AI refers to autonomous AI systems that can perform tasks independently to achieve specific goals.

  • How does agentic AI differ from generative AI?

    Generative AI focuses on creating content, while agentic AI can operate workflows autonomously to meet defined business outcomes.

  • What frameworks are useful for developing agentic AI applications?

    Frameworks like F Data, Microsoft Autogen, and Lang Chain are effective for building agentic AI applications.

  • Can agentic AI integrate with external tools?

    Yes, agentic AI can integrate with various external tools and APIs to enhance its capabilities.

  • What examples of agentic AI applications are discussed?

    Examples include financial analysis applications that compare stock performance and suggest optimal investments.

  • What role do LLM models play in agentic AI?

    LLM models are often integrated into agentic AI systems to process and analyze data from various sources.

Ver mais resumos de vídeos

Obtenha acesso instantâneo a resumos gratuitos de vídeos do YouTube com tecnologia de IA!
Legendas
en
Rolagem automática:
  • 00:00:00
    hello all my name is Krish naak and
  • 00:00:02
    welcome to my YouTube channel so guys as
  • 00:00:05
    you all know that we are towards the end
  • 00:00:07
    of 2024 uh there are some few more days
  • 00:00:10
    to probably uh you know complete 2024
  • 00:00:13
    and we are moving towards 2025 and if I
  • 00:00:16
    talk about or give you a brief summary
  • 00:00:18
    about this specific year with respect to
  • 00:00:20
    the development of generative AI right
  • 00:00:22
    in the field of generative AI we saw
  • 00:00:24
    various llm models we saw Tech Giants
  • 00:00:26
    competing between themselves to probably
  • 00:00:29
    create the LM models in terms of uh you
  • 00:00:32
    know applications in terms of content
  • 00:00:34
    generation and many more uh we as a
  • 00:00:37
    developer obviously we like we love to
  • 00:00:39
    learn all those things so we also
  • 00:00:41
    thought of using different different
  • 00:00:43
    Frameworks and how we could probably
  • 00:00:45
    integrate this kind of llm models and
  • 00:00:47
    create some amazing custom chat Bots I
  • 00:00:49
    think in this entire year we have
  • 00:00:51
    specifically done this if I talk
  • 00:00:53
    probably talk from the mid of this
  • 00:00:54
    particular year there is something
  • 00:00:56
    called as AI agents or agentic AI that
  • 00:00:58
    have actually come you know uh companies
  • 00:01:01
    have started talking more about multi-
  • 00:01:03
    aai agents you know or we also say
  • 00:01:05
    agentic AI you know so in this specific
  • 00:01:08
    video uh I will be talking about what
  • 00:01:11
    exactly agentic AI is because the reason
  • 00:01:13
    of making this specific video is that
  • 00:01:15
    what I feel uh now in this particular
  • 00:01:17
    field of generative AI if I probably see
  • 00:01:19
    in the upcoming years uh there are many
  • 00:01:22
    many people who are going to work on
  • 00:01:24
    this specific skill set that is agentic
  • 00:01:26
    AI now what exactly is agentic AI how is
  • 00:01:29
    is it different from generative AI I
  • 00:01:31
    will be covering each and everything in
  • 00:01:33
    this specific video along with this uh I
  • 00:01:36
    will also be talking about which all
  • 00:01:38
    Frameworks you can specifically use uh
  • 00:01:41
    to probably create this agentic AI
  • 00:01:43
    applications you know and uh if you
  • 00:01:46
    probably see that while we go in 2025 I
  • 00:01:49
    am going to probably cover many many
  • 00:01:51
    many Frameworks and build amazing end to
  • 00:01:53
    endend use cases with the help of
  • 00:01:55
    agentic AI application and I'm also very
  • 00:01:57
    near to launch my own mock interview
  • 00:02:00
    platform initially we created that as a
  • 00:02:02
    generative AI application now we also
  • 00:02:04
    converting that into an agentic
  • 00:02:05
    application now to understand all these
  • 00:02:08
    things uh first of all let me just go
  • 00:02:10
    ahead and share my screen and here uh in
  • 00:02:13
    this particular video I will be making
  • 00:02:15
    you understand about what exactly
  • 00:02:16
    agentic AI is okay so uh till now
  • 00:02:20
    whenever we used to probably talk about
  • 00:02:22
    generative AI you know uh we used to
  • 00:02:24
    probably say that hey we have some kind
  • 00:02:26
    of models which will be able to help us
  • 00:02:29
    create content
  • 00:02:31
    right now here we are specifically using
  • 00:02:35
    any kind of llm models right so let's
  • 00:02:37
    say open a llm models or open source llm
  • 00:02:40
    models Lama 3 different different llm
  • 00:02:43
    models so if I probably consider these
  • 00:02:44
    llm models our main task was that we
  • 00:02:48
    used to query let's say a user is
  • 00:02:53
    querying user used to put some kind of
  • 00:02:56
    query and then what we used to do is
  • 00:02:58
    that uh along with this llm model we
  • 00:03:00
    also need to write a specific prompt
  • 00:03:03
    like how this llm model needs to behave
  • 00:03:06
    let's say I may probably say that hey
  • 00:03:07
    you need to probably act like a author
  • 00:03:10
    and you need to probably create a poem
  • 00:03:12
    on this specific query whatever query we
  • 00:03:15
    are trying to ask over here or the user
  • 00:03:17
    is trying to ask and then based on this
  • 00:03:19
    we used to specifically get some kind of
  • 00:03:22
    output right so if we see in this
  • 00:03:25
    particular flow the reason we used to
  • 00:03:28
    say this as a generative AI because at
  • 00:03:29
    at the end of the day it is creating or
  • 00:03:32
    it is generating it is generating
  • 00:03:35
    content right it is generating content
  • 00:03:38
    and this is what is all about generative
  • 00:03:40
    AI application you can probably finetune
  • 00:03:42
    the prompt you can make this llm model
  • 00:03:45
    behave in a better way you can probably
  • 00:03:46
    create custom chat Bots uh we can modify
  • 00:03:49
    this prompt we can uh we also say it as
  • 00:03:52
    prompt finetuning you can probably make
  • 00:03:55
    sure to write a right kind of prompt so
  • 00:03:57
    that you will be able to get the right
  • 00:03:58
    output but if I talk with respect to
  • 00:04:02
    agentic AI application okay so agentic
  • 00:04:04
    AI
  • 00:04:05
    application now whenever we talk about
  • 00:04:08
    agentic AI or let's talk about this
  • 00:04:11
    agentic AI these are nothing but these
  • 00:04:14
    are
  • 00:04:15
    autonomous
  • 00:04:17
    autonomous AI
  • 00:04:19
    system okay autonomous AI system now
  • 00:04:23
    let's say uh when we were specifically
  • 00:04:26
    talking with respect to generative AI
  • 00:04:28
    application there were also some
  • 00:04:29
    applications that we used right like
  • 00:04:31
    let's say rag right rag rag application
  • 00:04:35
    I hope everybody has heard about it now
  • 00:04:37
    what rag application basically does is
  • 00:04:39
    that you know these llm models uh are
  • 00:04:42
    not currently trained with the recent
  • 00:04:44
    data you know it is regularly trained
  • 00:04:47
    every 6 months every 8 months every year
  • 00:04:50
    right but when I probably say today what
  • 00:04:52
    we have this recent data in the internet
  • 00:04:54
    it has not been trained with that you
  • 00:04:56
    know if I probably ask any llm model hey
  • 00:04:58
    what is the current news that I have
  • 00:05:00
    right can you probably talk about the
  • 00:05:01
    top AI news five top AI news then this
  • 00:05:05
    llm model will not be able to answer it
  • 00:05:07
    so what happens is that whenever a user
  • 00:05:09
    try to query you know this llm model
  • 00:05:11
    needs to also have access to something
  • 00:05:15
    like an external external tool okay or
  • 00:05:18
    external Source I can probably say this
  • 00:05:21
    as external
  • 00:05:23
    Source right now right now this
  • 00:05:26
    particular exis is not there so what we
  • 00:05:28
    do is that if you probably use some
  • 00:05:30
    Frameworks like Lang chain and all we
  • 00:05:32
    have something called as tools right one
  • 00:05:35
    of the tool can be Doug dug go search I
  • 00:05:38
    hope everybody has heard about this this
  • 00:05:40
    basically helps you to do the Google
  • 00:05:43
    search or uh do the search uh with
  • 00:05:46
    respect to the recent news that are
  • 00:05:47
    there in the internet okay you also have
  • 00:05:49
    some other tools like let's say I want
  • 00:05:51
    to specifically use uh Wikipedia search
  • 00:05:54
    right so I can have Wikipedia tool for
  • 00:05:58
    this and langin probably every tool is
  • 00:06:00
    basically created if you also want to
  • 00:06:02
    probably explore some other tool like RF
  • 00:06:05
    so that is also available so Lang chain
  • 00:06:07
    what it did is that it came up with this
  • 00:06:09
    kind of various tools which you can
  • 00:06:11
    probably integrate along with the llm
  • 00:06:13
    applications to generate some kind of
  • 00:06:15
    content okay but see you may be thinking
  • 00:06:19
    if this external tools are there why
  • 00:06:21
    can't we probably create autonomous AI
  • 00:06:23
    system that can be able to perform this
  • 00:06:25
    their own task now see guys over here
  • 00:06:28
    when I I talk about autonomous AI system
  • 00:06:30
    you know with respect to agent AI there
  • 00:06:33
    is a specific goal to achieve right and
  • 00:06:36
    while achieving this particular goal
  • 00:06:38
    this all AI system will be working
  • 00:06:41
    independently without any human
  • 00:06:43
    intervention okay to achieve this
  • 00:06:45
    specific goal okay now in this
  • 00:06:47
    particular case if let's say that I
  • 00:06:50
    require one tool so I may integrate if I
  • 00:06:52
    require another tool that may also get
  • 00:06:54
    integrated if I may uh probably use one
  • 00:06:56
    more tool it may also get integrated
  • 00:06:58
    right so those kind of tools you can
  • 00:07:00
    manually integrate it if you have also
  • 00:07:03
    used Lang graph okay Lang graph is one
  • 00:07:06
    amazing uh module that has been coming
  • 00:07:09
    by Lang chain you know and here also you
  • 00:07:10
    can probably create agentic AI
  • 00:07:12
    applications right uh this is also one
  • 00:07:15
    of the Frameworks that you can
  • 00:07:16
    specifically use but I hope you're
  • 00:07:18
    understanding what is the basic
  • 00:07:19
    difference between generative Ai and
  • 00:07:20
    agentic AI in generative AI the main aim
  • 00:07:24
    is just to create the content right if
  • 00:07:26
    you are trying to make this applic ation
  • 00:07:30
    uh have access to an external Source
  • 00:07:33
    right here the workflow becomes quite
  • 00:07:36
    complex because I have to keep on adding
  • 00:07:38
    one or the other tool one or the other
  • 00:07:40
    tool and there is no such end goal over
  • 00:07:42
    here the main aim of this llm models is
  • 00:07:45
    just to create content now similarly
  • 00:07:47
    what we have actually done this agentic
  • 00:07:50
    AI that is basically coming recently
  • 00:07:52
    right it is really becoming popular many
  • 00:07:54
    cosos are actually talking about it uh
  • 00:07:56
    even Microsoft coo has said that hey
  • 00:07:58
    this is probably going to replace the
  • 00:08:01
    SAS products you know and it is it may
  • 00:08:04
    be true guys because the application is
  • 00:08:06
    quite amazing so as I said when compared
  • 00:08:09
    to generative AI where the main name is
  • 00:08:11
    not to probably create content as the
  • 00:08:14
    goal here the goal will be based on the
  • 00:08:16
    business outcome that you really want to
  • 00:08:18
    do right business outcome business
  • 00:08:22
    outcome and the best thing will be that
  • 00:08:24
    this autonomous AI agents can have a
  • 00:08:28
    very complex
  • 00:08:30
    workflow very complex workflow right and
  • 00:08:36
    this complex workflow will get executed
  • 00:08:39
    independently to achieve this goal right
  • 00:08:43
    and parall as this task is being
  • 00:08:46
    probably completed right if it is
  • 00:08:48
    getting completed this autonomous AI
  • 00:08:51
    system can also fine tune themselves to
  • 00:08:54
    perform much more better to perform much
  • 00:08:59
    more better and this was not there in
  • 00:09:02
    generative AI applications right so that
  • 00:09:04
    is the reason there are so many
  • 00:09:05
    Frameworks that are coming there's some
  • 00:09:07
    Frameworks like Lang Lang graph there is
  • 00:09:09
    framework like langlow there is a
  • 00:09:11
    framework like fi data I'll also show
  • 00:09:13
    you one amazing end to end examples
  • 00:09:15
    which I have probably implemented for
  • 00:09:17
    you okay now this is with respect to the
  • 00:09:19
    agentic AI right now let me talk about
  • 00:09:22
    some of the examples so that you get a
  • 00:09:23
    complete idea like how this autonomous
  • 00:09:25
    AI system works let's say that my task
  • 00:09:27
    is probably to create
  • 00:09:29
    a custom bot okay custom bot and this
  • 00:09:34
    bot has uh this bot is nothing but it is
  • 00:09:37
    an agentic AI application okay agentic
  • 00:09:41
    AI application which probably interacts
  • 00:09:44
    with many many things okay let's say my
  • 00:09:47
    main goal is that this custom bot main
  • 00:09:51
    goal is that whenever I ask any
  • 00:09:54
    query I ask any query let's say that I
  • 00:09:56
    ask this bot related something related
  • 00:09:59
    to finance I'll say that hey I have
  • 00:10:02
    $1,000 okay I want to probably buy some
  • 00:10:06
    stocks buy some stocks and let's say
  • 00:10:10
    that this is this particular stocks I
  • 00:10:12
    want to sell it in 5 days sell in 5 days
  • 00:10:17
    and I want a minimal profit of 50% let's
  • 00:10:19
    say I'm just putting some context this
  • 00:10:22
    is my goal that I really want to do I
  • 00:10:25
    have probably kept a goal right now as
  • 00:10:28
    soon as I do this
  • 00:10:30
    now what this custom bot is basically
  • 00:10:32
    going to do which is like an agentic AI
  • 00:10:35
    this is going to create a complex
  • 00:10:37
    workflow now the workflow let's say we
  • 00:10:40
    have created a workflow this should be
  • 00:10:41
    dependent on your uh Finance news let's
  • 00:10:45
    say for finance news we can use YT YT
  • 00:10:48
    Finance right so YT Finance
  • 00:10:51
    functionality is also there YT Finance
  • 00:10:53
    actually helps you to get all the stock
  • 00:10:55
    details how the performance is like all
  • 00:10:58
    the information that is probably there
  • 00:10:59
    in the website right and this custom bot
  • 00:11:02
    is just like an llm model this llm model
  • 00:11:05
    can take this particular information and
  • 00:11:07
    start working on it now along with this
  • 00:11:09
    I may also require some other
  • 00:11:11
    information what is the recent current
  • 00:11:14
    news recent current news right so recent
  • 00:11:18
    current news also so this referral
  • 00:11:20
    should also be going now my workflow
  • 00:11:22
    will look something like this first it
  • 00:11:24
    goes and picks up all the detail related
  • 00:11:25
    to finance news let's say that uh I just
  • 00:11:28
    written by buy some stocks sell it in 5
  • 00:11:30
    days get a 50% or let's say my query is
  • 00:11:33
    that hey just let me know how is this
  • 00:11:36
    Tesla stock okay how is this Tesla stock
  • 00:11:39
    let's say uh I have something like uh
  • 00:11:41
    okay I want to probably check what is
  • 00:11:43
    this Tesla stock let me just uh change
  • 00:11:46
    my pen okay okay perfect so let's say
  • 00:11:51
    that I have a stock I want to probably
  • 00:11:53
    compare and say that hey compare between
  • 00:11:55
    Tesla and Nvidia and suggest me which
  • 00:11:58
    one should I bu right so this if this is
  • 00:12:01
    my question to my custom bot or to this
  • 00:12:04
    agent AI application so the workflow
  • 00:12:06
    will get started so my workflow will
  • 00:12:08
    probably be in such a way that first of
  • 00:12:10
    all I will go ahead and refer YT Finance
  • 00:12:12
    okay I'll get the information about both
  • 00:12:14
    the stock I will probably Compare the
  • 00:12:16
    numbers the llm model will already
  • 00:12:17
    Compare the numbers and I'll provide the
  • 00:12:19
    information over there then my next flow
  • 00:12:21
    will be that hey just go and see the
  • 00:12:22
    current news like which what is a
  • 00:12:25
    positive news what is a positive
  • 00:12:26
    sentiment news that are probably there
  • 00:12:28
    so I may probably go and see the top
  • 00:12:30
    recent news right after this I may
  • 00:12:32
    probably say that hey what kind of
  • 00:12:34
    recent development is specifically
  • 00:12:35
    happening right so this will also refer
  • 00:12:38
    to some other news API right which is
  • 00:12:40
    just acting like a tool right so this
  • 00:12:43
    when I'm saying this is a tool this is
  • 00:12:45
    basically an agentic AI application it
  • 00:12:47
    is an autonomous AI Auto autonomous AI
  • 00:12:51
    so this is a separate AI agent this is a
  • 00:12:53
    separate AI agent this is a separate AI
  • 00:12:55
    agent now this all executing together
  • 00:12:58
    each and every task is basically handled
  • 00:13:01
    by a separate autonomous AI agent right
  • 00:13:04
    and they are working together they are
  • 00:13:06
    executing this in a workflow right and
  • 00:13:09
    finally after interacting with this you
  • 00:13:11
    may get an output right you may get an
  • 00:13:14
    output or the kind of suggestion that
  • 00:13:15
    you probably looking for because what
  • 00:13:17
    kind of question have you asked hey
  • 00:13:18
    should I probably I I'll just say that
  • 00:13:21
    hey go ahead and compare Tesla versus
  • 00:13:22
    Nvidia stock and tell me which one
  • 00:13:24
    should I buy so after doing all this
  • 00:13:26
    after executing this entire workflow we
  • 00:13:29
    will be getting a very good summary
  • 00:13:31
    saying that hey what you really need to
  • 00:13:32
    buy right this was not possible with the
  • 00:13:35
    help of generative AI application
  • 00:13:36
    because here the main aim is just to
  • 00:13:39
    create content but if I talk about
  • 00:13:41
    agentic AI application this has become
  • 00:13:43
    much more smarter it is working
  • 00:13:45
    autonomously and you can create any kind
  • 00:13:48
    of complex workflow let's say you want
  • 00:13:50
    to integrate multiple tools you want to
  • 00:13:52
    integrate multiple AI autonomous agents
  • 00:13:54
    you are able to do it and that is what
  • 00:13:56
    is all about AI agents or agentic a
  • 00:13:59
    right now what I am actually going to
  • 00:14:02
    show you is that what all Frameworks you
  • 00:14:04
    can specifically use and what I feel
  • 00:14:07
    that in the upcoming years this
  • 00:14:09
    Frameworks will be quite amazing because
  • 00:14:11
    uh if I talk about I built my first
  • 00:14:14
    agentic AI application 6 months back and
  • 00:14:16
    now more and more amazing Frameworks are
  • 00:14:18
    specifically coming up now these all
  • 00:14:20
    Frameworks are open source and I'll also
  • 00:14:22
    show you one end to endend example as we
  • 00:14:24
    go ahead so let's go ahead and let's
  • 00:14:26
    discuss about different different
  • 00:14:27
    Frameworks but I hope you got an amazing
  • 00:14:30
    example to to find out the differences
  • 00:14:32
    between generative Ai and agentic AI
  • 00:14:34
    right so guys now let us go ahead and
  • 00:14:36
    probably discuss about some of the
  • 00:14:38
    amazing Frameworks that are right now in
  • 00:14:40
    there in the market which will actually
  • 00:14:42
    help you to build about agent TK
  • 00:14:43
    application once we specifically discuss
  • 00:14:46
    about this I'm also going to show you
  • 00:14:47
    one amazing end to end uh example where
  • 00:14:49
    I've actually created a financial
  • 00:14:51
    analyst which will actually help you to
  • 00:14:53
    decide which stocks you can currently
  • 00:14:55
    buy based on the recent information that
  • 00:14:58
    we have have related to that particular
  • 00:15:00
    stock price okay so here is one of the
  • 00:15:03
    amazing framework which is called as F
  • 00:15:04
    data now in that in this F data you will
  • 00:15:07
    be able to create different different
  • 00:15:08
    agents like legal agent financial
  • 00:15:10
    analysis construction analyst marketing
  • 00:15:13
    agent and these are like uh agentic AI
  • 00:15:15
    application you can execute them in a
  • 00:15:17
    specific workflow uh right now this is
  • 00:15:20
    really really open source and uh here
  • 00:15:23
    many many people are many many companies
  • 00:15:25
    are specifically using it our end to end
  • 00:15:27
    examples what we are basically going to
  • 00:15:29
    do is that we going to use this and
  • 00:15:31
    probably create that okay now uh you the
  • 00:15:33
    best thing about this is that you will
  • 00:15:35
    be able to integrate with any llm models
  • 00:15:37
    that you specifically want from grock to
  • 00:15:40
    Google Gemini to anthropic to cloudy to
  • 00:15:43
    AWS to openi to Mistral anything that
  • 00:15:46
    you specifically want okay then along
  • 00:15:49
    with this you can also add any number of
  • 00:15:51
    tools that you want right so it enable
  • 00:15:53
    agents to integrates and interact with
  • 00:15:55
    external system right so the autonomous
  • 00:15:58
    AI system as said right hey let's
  • 00:16:00
    probably if I ask hey which stock should
  • 00:16:01
    I buy between Tesla or this I will
  • 00:16:04
    probably go ahead and create an AI agent
  • 00:16:06
    that AI agent will start uh you know
  • 00:16:08
    communicating with the external uh apis
  • 00:16:11
    News Channel or anything that you want
  • 00:16:13
    right even you have options to integrate
  • 00:16:15
    with payment Gateway and all right so
  • 00:16:17
    that is the most amazing thing and is
  • 00:16:19
    pretty much simple once I probably show
  • 00:16:21
    you you can definitely have a look okay
  • 00:16:23
    the next common and the famous uh um you
  • 00:16:26
    know product or agentic AI framework
  • 00:16:28
    work is nothing but Microsoft autogen
  • 00:16:31
    right so this autogen is also coming
  • 00:16:34
    like recently it is in the Talk of the
  • 00:16:36
    Town and it it is again open source and
  • 00:16:38
    it basically helps you to create a
  • 00:16:40
    framework for agentic AI application
  • 00:16:43
    along with this some more Frameworks
  • 00:16:45
    that you can probably explore is
  • 00:16:46
    something called as langlow now Lang
  • 00:16:48
    flow I've already uploaded some videos
  • 00:16:49
    in my channel this is a complete no code
  • 00:16:53
    tool probably to develop your agentic AI
  • 00:16:55
    applications which uh where you can also
  • 00:16:58
    design complex workflows so uh yes you
  • 00:17:01
    can probably do this and this is more
  • 00:17:03
    like a drag and drop automatically your
  • 00:17:05
    code your endpoint code will also be
  • 00:17:07
    provided and you should also be able to
  • 00:17:09
    deploy this in an amazing way so this is
  • 00:17:11
    just one example here you can see uh
  • 00:17:13
    user is probably creating an agent this
  • 00:17:16
    agent is probably doing some kind of
  • 00:17:18
    research uh the research is basically
  • 00:17:20
    done based on the content research uh
  • 00:17:22
    here you have different different API
  • 00:17:24
    integration you can also go ahead and
  • 00:17:25
    use your own prompting and many more so
  • 00:17:28
    just to show you example let's say that
  • 00:17:29
    I've created this simple agent over here
  • 00:17:32
    and this is my entire uh you know
  • 00:17:34
    workflow here you can see that I've I've
  • 00:17:36
    used multiple agents URL calculator chat
  • 00:17:39
    input dugdug go search and here I've
  • 00:17:42
    used an agent with open Ai and finally
  • 00:17:44
    I'm able to get the output once you
  • 00:17:46
    probably create this you can also get
  • 00:17:48
    your endtoend code over here so you just
  • 00:17:51
    need to probably get your python API
  • 00:17:52
    JavaScript API or python code whatever
  • 00:17:55
    you like and you can integrate it and
  • 00:17:56
    you can probably uh take it to that that
  • 00:17:59
    entire Next Level okay then uh you have
  • 00:18:02
    also one more amazing uh framework that
  • 00:18:05
    is called as Lang graph and with respect
  • 00:18:07
    to Lang graph here also you can create
  • 00:18:09
    complex workflows uh if I consider Lang
  • 00:18:11
    graph more complex workflows you can
  • 00:18:13
    specifically create if I talk about data
  • 00:18:15
    stack slang flow here you able to create
  • 00:18:17
    complex workflow you can you can also
  • 00:18:19
    use multiple agents as of your
  • 00:18:21
    requirement multiple models multiple
  • 00:18:23
    data preprocessing techniques multiple
  • 00:18:25
    prompt techniques and all um here you
  • 00:18:28
    can use Amazon Bedrock anthropic open AI
  • 00:18:30
    cair Google generative AI Gro anything
  • 00:18:33
    that you specifically want uh right now
  • 00:18:35
    this is probably my favorite because
  • 00:18:37
    with the help of this I'm able to create
  • 00:18:39
    an application much more easier but yes
  • 00:18:41
    my next year plan is to probably cover
  • 00:18:43
    all this kind of amazing tools and
  • 00:18:45
    develop some amazing end to end agent
  • 00:18:49
    applications now let's go ahead and
  • 00:18:51
    probably see one example I'll be using F
  • 00:18:53
    data and I'll show you that how you can
  • 00:18:55
    probably go ahead and create your own
  • 00:18:57
    agent application
  • 00:18:59
    so guys so here is one amazing endtoend
  • 00:19:02
    agentic AI application uh where I'm
  • 00:19:05
    specifically using open AI API key F API
  • 00:19:08
    key here my main name is that I've
  • 00:19:10
    created an agent which is called as web
  • 00:19:12
    search agent and this main work is
  • 00:19:14
    basically to interact with external
  • 00:19:17
    thirdparty sources with respect to the
  • 00:19:19
    recent news that we have so here you can
  • 00:19:21
    see dougd Source search it is basically
  • 00:19:23
    using the model it is using is open AI
  • 00:19:25
    chat this is my web search agent along
  • 00:19:27
    with that I also have a financial agent
  • 00:19:29
    which is interacting with the Y Finance
  • 00:19:30
    tool to get the stock market prices
  • 00:19:33
    details and all and uh we can combine
  • 00:19:35
    both the specific agent to probably
  • 00:19:38
    create my entire playground where I will
  • 00:19:40
    be able to create it in as a custom
  • 00:19:42
    chatbot okay so here uh one amazing
  • 00:19:45
    thing is that first of all we are
  • 00:19:46
    creating an agent and inside that agent
  • 00:19:48
    we can have any number of tools and
  • 00:19:50
    model integration similarly over here
  • 00:19:52
    this is my financial agent but
  • 00:19:54
    internally you'll be able to see that it
  • 00:19:56
    is interacting with tools like why
  • 00:19:58
    Finance tool and it is using a model
  • 00:19:59
    like open AI chat right and similarly we
  • 00:20:01
    can probably use different different
  • 00:20:03
    combination now this kind of projects I
  • 00:20:06
    will also be coming up uh probably
  • 00:20:08
    implementing everything from scratch but
  • 00:20:10
    here I actually used this F data now
  • 00:20:13
    this is running uh and here uh let's go
  • 00:20:15
    ahead and probably check this in my um f
  • 00:20:19
    data once you need to probably log in
  • 00:20:20
    you'll be able to check it out in in the
  • 00:20:22
    form of demo over
  • 00:20:24
    here so guys now I am inside this
  • 00:20:27
    particular playground of data. app all
  • 00:20:29
    you have to do is probably login and
  • 00:20:31
    right now you can also see that my code
  • 00:20:33
    is running over here all I have to do is
  • 00:20:35
    that I have to probably connect to an
  • 00:20:36
    endpoint so the current my endpoint is
  • 00:20:38
    basically running in Local Host uh 777
  • 00:20:42
    okay and uh here you'll I'll just see
  • 00:20:45
    that I'll just go ahead and ask hey what
  • 00:20:46
    is your special skill okay and uh here
  • 00:20:49
    you can probably see my answer right
  • 00:20:52
    whether you need financial analysis St a
  • 00:20:54
    recommendation or anything you'll be
  • 00:20:55
    there to help right so now let's go
  • 00:20:57
    ahead and ask some more questions uh
  • 00:20:59
    I'll say hey can you
  • 00:21:01
    provide
  • 00:21:03
    me uh can you suggest me which stock to
  • 00:21:10
    buy um is it
  • 00:21:15
    between Tesla and Nvidia okay so let's
  • 00:21:20
    say that this is my question okay and
  • 00:21:23
    now here you can see that entire things
  • 00:21:25
    are basically running yeah so I'm able
  • 00:21:27
    to get the answer see over here Nvidia
  • 00:21:29
    so and so this is the price overall hold
  • 00:21:32
    based on the data analyze strong
  • 00:21:34
    Contender Manus uh Nvidia shows strong
  • 00:21:38
    buy Tesla also shows strong buy but the
  • 00:21:41
    number is 12 or six buy 48 hold 4 so it
  • 00:21:45
    appears to be a stronger Contender with
  • 00:21:47
    an un strong buy recommendation this so
  • 00:21:50
    here what is basically happening is that
  • 00:21:51
    internally agents is basically getting
  • 00:21:53
    called all the information from the
  • 00:21:55
    stocks is basically getting up over here
  • 00:21:57
    let's say that if I want to probably go
  • 00:21:59
    ahead and click on this I will also be
  • 00:22:00
    able to get maximum information as we
  • 00:22:02
    can so it is probably taking all this
  • 00:22:04
    information and based on that it is
  • 00:22:06
    making this kind of decision so why we
  • 00:22:09
    say this as agentic AI because
  • 00:22:10
    internally all these agents has been
  • 00:22:12
    calling this is basically interacting
  • 00:22:14
    with this kind of uh uh external tools
  • 00:22:18
    it is working in the form of a workflow
  • 00:22:20
    right so I hope you like this particular
  • 00:22:22
    video as we go ahead I've also created
  • 00:22:24
    this kind of projects it'll be coming in
  • 00:22:26
    my YouTube channel how you can probably
  • 00:22:27
    go ahead and start from Basics so yeah
  • 00:22:29
    this was it from my side I hope you like
  • 00:22:31
    this particular video I'll see you in
  • 00:22:32
    the next video itself thank you have a
  • 00:22:34
    great day
Etiquetas
  • Agentic AI
  • Generative AI
  • AI Frameworks
  • LLM Models
  • Autonomous Systems
  • Finance
  • Chatbots
  • Workflow Automation
  • Data Analysis
  • Artificial Intelligence