Custom GPT Tools Tutorial | Get Real-time Weather of Any Location with API | SeaChat by Seasalt.ai

00:12:37
https://www.youtube.com/watch?v=2C2IOabHHFc

Resumo

TLDRIn this advanced tutorial, Shen Yao from se AI explains how to use custom GPD tools within CChat to perform complex queries, specifically focusing on obtaining weather data through an API. The session builds on previous knowledge of simple queries and ventures into assembling API endpoints by combining fixed and dynamic parameters such as timezone, latitude, and longitude. An explanation of 'temperature 2m' is provided as temperature at 2 meters above ground. Shen demonstrates assembling a query that converts location names into latitude and longitude using the description fields in the tool setup, and shares a step-by-step guide on setting fixed and dynamic parameters. The demonstration includes real-time testing and debugging techniques to ensure accuracy of results, such as ensuring API responses are correctly interpreted for weather forecasts. This method is applicable for other complex API queries beyond weather data.

Conclusões

  • 🌦️ Learn to fetch weather data using GPD tools.
  • 🛠️ Create custom GPD tools in CChat.
  • 🗺️ Convert location names to latitude and longitude.
  • 📊 Assemble complex API queries with fixed and dynamic parameters.
  • 🧩 Explanation of 'temperature 2m'.
  • 🔄 Debug API calls for accurate weather forecasts.
  • 🌍 Manage timezones and geographical data in queries.
  • 🔍 Real-time testing of API responses.
  • 👨‍💻 Techniques applicable for other complex queries.
  • 🔒 Secure API handling and data fetching.

Linha do tempo

  • 00:00:00 - 00:05:00

    In this advanced tutorial, Shen Yao, a product manager at SE AI, demonstrates using custom GPT tools to perform complex weather queries. In the previous simpler tutorial, users learned to retrieve animal images using specific APIs. This session progresses to accessing weather information using an open weather API. Shen walks through creating a new custom GPT tool by assembling required endpoint URLs and parameters such as latitude, longitude, timezone, and forecast period. A critical element involves understanding parameter meanings, like 'temperature 2m', which refers to measuring temperatures at two meters above ground level. He illustrates how to set these API parameters, using Seattle's forecast as an example, and highlights the integration process for a comprehensive weather data retrieval setup.

  • 00:05:00 - 00:12:37

    Shen continues by configuring dynamic and fixed parameters for the API, such as location-based latitude and longitude, which can be converted directly from user input within the chat. After assembling the API call parameters, Shen demonstrates testing the API with different locations, such as Seattle and Chicago, observing real-time weather data retrieved via the configured GPT tool. Debugging features are showcased to illustrate how responses are constructed and verified. The session concludes with confirming the accuracy and functionality of converting geographical locations to latitudinal and longitudinal data for weather predictions. He emphasizes the utility of these advanced custom queries in enhancing user interactions with weather-related tasks in cchat.

Mapa mental

Mind Map

Perguntas frequentes

  • What is the main focus of this tutorial?

    The tutorial focuses on using customer GPD tools to perform complex queries like fetching weather information using APIs.

  • Who is presenting the tutorial?

    The tutorial is presented by Shen Yao, a product manager at se AI.

  • What specific task is being demonstrated?

    Creating a custom GPD tool to fetch weather forecasts using a weather API is demonstrated.

  • What tools are used in the tutorial?

    Custom GPD tools and the weather API in conjunction with CChat are used.

  • How are API parameters handled in the tutorial?

    The API parameters are set by defining fixed values and dynamic parameters such as latitude and longitude.

  • What is the explanation for 'temperature 2m'?

    'Temperature 2m' refers to the temperature measured at 2 meters above the surface.

  • How is location converted for API use?

    Location names are converted to latitude and longitude using the instructions given in the GPD tool.

  • Can the tutorial's strategies be used for other APIs?

    Yes, the strategies can be adapted for different APIs requiring parameter assembly.

  • What type of request is demonstrated for obtaining weather data?

    A GET request is demonstrated for obtaining weather data.

  • What debugging technique is shown at the end?

    The video shows using debug mode to view API call outputs and verify location conversions.

Ver mais resumos de vídeos

Obtenha acesso instantâneo a resumos gratuitos de vídeos do YouTube com tecnologia de IA!
Legendas
en
Rolagem automática:
  • 00:00:01
    hello everybody my name is Shen Yao I'm
  • 00:00:03
    a product manager at se AI uh today
  • 00:00:06
    we're going to go into some Advanced
  • 00:00:08
    tutorial of using customer GPD tools uh
  • 00:00:11
    to um perform some complex queries using
  • 00:00:14
    cchat so uh last time in my uh simple
  • 00:00:17
    tutorial I have shown you the customer
  • 00:00:20
    GPD tools over here and then we are able
  • 00:00:22
    to uh get some very very simple get a
  • 00:00:25
    dog picture or get a fox picture by
  • 00:00:27
    calling this API right dog API I over
  • 00:00:30
    here random dock um image and then um
  • 00:00:34
    random fox image over here right so uh
  • 00:00:37
    in this session I'm going to move into
  • 00:00:38
    the more advanced ones of getting
  • 00:00:41
    getting the weather right so I got a
  • 00:00:42
    free and open medial weather apepi over
  • 00:00:46
    here uh it's kind of complicated because
  • 00:00:50
    um because you have to assemble all
  • 00:00:52
    kinds of different parameters right but
  • 00:00:54
    uh let's give it a try so first let's go
  • 00:00:57
    to the documentation and then let's see
  • 00:00:59
    um
  • 00:01:00
    how to assemble this uh weather API so
  • 00:01:03
    what I'm going to do is that I'm going
  • 00:01:04
    to create a new customer GPD tool right
  • 00:01:07
    so if I go back over here you see U uh I
  • 00:01:10
    can add up to five customer GPD tools
  • 00:01:12
    over here and then enable this I will
  • 00:01:14
    say get weather uh API right so this is
  • 00:01:19
    essentially a get request and then we're
  • 00:01:22
    going to get endpoint URL and then we're
  • 00:01:24
    going to assemble all all of those
  • 00:01:25
    parameters right so how this one works
  • 00:01:28
    is this right so we have a
  • 00:01:30
    from this API
  • 00:01:33
    page uh eventually we're going to
  • 00:01:34
    assemble a final URL something
  • 00:01:38
    like something like um somewhere in this
  • 00:01:43
    uh in this place right so uh uh let's
  • 00:01:47
    see module update URL over here oh okay
  • 00:01:50
    I we actually got it over here right so
  • 00:01:53
    this is a this is the URL over here so
  • 00:01:56
    uh what we're going to do is this right
  • 00:01:59
    so I have end point URL which is all the
  • 00:02:03
    way to to up here forecast right so I'm
  • 00:02:06
    going to I'm going to change this right
  • 00:02:09
    and then we're going to say this is a
  • 00:02:10
    get weather
  • 00:02:11
    forecast uh get weather API right so the
  • 00:02:14
    description is that uh uh get weather
  • 00:02:18
    forecast of
  • 00:02:21
    forecast
  • 00:02:23
    um of a specific location right so
  • 00:02:28
    that's one and then you say okay uh from
  • 00:02:32
    here um I have a fixed latitude and a
  • 00:02:35
    longitude and then the time zone is not
  • 00:02:38
    a set up right since I'm a based in
  • 00:02:39
    Seattle so I'm going to choose America
  • 00:02:41
    Los Angeles T Zone and uh we're going to
  • 00:02:45
    do the for forecast of seven days and
  • 00:02:48
    then we're going to do the temperature
  • 00:02:50
    2m oh I don't really know what the
  • 00:02:52
    temperature 2m means um and we're going
  • 00:02:55
    to
  • 00:02:55
    say the prescription probability and
  • 00:02:58
    whether it's going to ring or not uh
  • 00:03:00
    since this is down in November and in
  • 00:03:02
    November I guess Seattle doesn't really
  • 00:03:04
    um snow that much so um let me ask a
  • 00:03:07
    Gemini first what is uh what is uh I got
  • 00:03:11
    a curious about this uh temperature 2m
  • 00:03:14
    right so uh what is 2m in uh in
  • 00:03:17
    temperature 2m over
  • 00:03:20
    here uh oh okay got it yeah so 2 2 m
  • 00:03:24
    actually means 2 m right so the
  • 00:03:25
    temperature measure the height of 2 m
  • 00:03:27
    above above the surface okay okay that's
  • 00:03:30
    clear enough right so so after we have
  • 00:03:33
    all of this and then we're going to
  • 00:03:34
    assemble uh I'm going to reload the
  • 00:03:37
    chart and this is actually the uh from
  • 00:03:40
    this location right from this location's
  • 00:03:42
    uh weather weather report over here and
  • 00:03:45
    then that's the uh that's the entire API
  • 00:03:47
    URL right so if I copy paste this and I
  • 00:03:50
    go back over here we already have the
  • 00:03:52
    Endo right so
  • 00:03:53
    um I already have this end point and
  • 00:03:56
    then we're going to assemble the rest of
  • 00:03:57
    this right latitude longitude and
  • 00:04:00
    already is seual uh so we're going to
  • 00:04:02
    walk backwards over here right from here
  • 00:04:05
    we got two pieces one is fixed value
  • 00:04:07
    parameter and then the other one is
  • 00:04:09
    dynamic variable parameter extracted by
  • 00:04:11
    the LM right so the fixed Val parameters
  • 00:04:15
    are something like uh the predefined
  • 00:04:17
    ones right so for instance we have
  • 00:04:18
    something called a time zone right so I
  • 00:04:20
    already know that U um I want
  • 00:04:24
    to I want to uh I want the the time zone
  • 00:04:27
    um specifically to Los Angeles so we're
  • 00:04:30
    going to say okay I have the query here
  • 00:04:32
    and my key here is called a time zone
  • 00:04:35
    and then I have this fixed as value it's
  • 00:04:37
    called America uh Los Angeles that's uh
  • 00:04:40
    that's uh that's one okay I'm going to
  • 00:04:42
    add another one which is um uh which is
  • 00:04:46
    this fixed parameter right so we got a
  • 00:04:49
    arly I got a a quy over here uh yes Cory
  • 00:04:54
    is on the URL parameter string and then
  • 00:04:56
    key key here is already right so we're
  • 00:04:59
    going to have already report of the
  • 00:05:02
    temperature at 2 m above the surface and
  • 00:05:04
    then I want the um precipitation
  • 00:05:07
    probability and whether it's going to
  • 00:05:08
    ring or not so those are the two fixed
  • 00:05:10
    value parameters prefilled by the user
  • 00:05:13
    and finally this is a more kind of a
  • 00:05:15
    challenging case we're going to assemble
  • 00:05:17
    the attitude latitude and longitude
  • 00:05:20
    right so so this is actually the dynamic
  • 00:05:22
    variable parameters here determined by
  • 00:05:24
    the LM so we again have this a quiry
  • 00:05:28
    piece and then we got the string here
  • 00:05:30
    and the first one is uh latitude right
  • 00:05:33
    so I'm going to save the latitude and um
  • 00:05:38
    the default value is just a 522 52 right
  • 00:05:41
    so um and then we're going to say this
  • 00:05:44
    is required right because the API
  • 00:05:46
    requires this so the description here is
  • 00:05:48
    a little bit tricky because no one is
  • 00:05:49
    going to uh you know enter the uh
  • 00:05:51
    latitude and longitude into the into a
  • 00:05:54
    chat window right so what's what we're
  • 00:05:55
    going to say is this uh I'm going to say
  • 00:05:58
    um I would to say
  • 00:06:00
    location uh location
  • 00:06:03
    of the
  • 00:06:04
    weather I would say
  • 00:06:07
    latitude of the location right and then
  • 00:06:10
    I'm going to say um convert location to
  • 00:06:15
    Latitude right so that's one and the
  • 00:06:17
    next part is longitude longitude right
  • 00:06:20
    so I I got a quy over here and I got a
  • 00:06:23
    string over here the key is longitude
  • 00:06:25
    longitude and that's
  • 00:06:28
    13.41% go out of the earth so I'm going
  • 00:06:31
    to copy exactly the same thing and then
  • 00:06:33
    for descriptions I'm going to uh repeat
  • 00:06:36
    this this is going to be the longitude
  • 00:06:38
    of the location conver the location to
  • 00:06:39
    longitude over here and then this is
  • 00:06:41
    required right now we have a sample this
  • 00:06:43
    API and I'm going to we uh C head
  • 00:06:47
    already
  • 00:06:48
    has um CAD already has this uh down
  • 00:06:53
    right assemble and I'm going to um run
  • 00:06:55
    run the test right so this looks a
  • 00:06:57
    little bit weird to me but we'll see
  • 00:07:00
    what happens right so I'm going to
  • 00:07:01
    submit and this says okay invited the
  • 00:07:04
    time zone so right America Los Angeles
  • 00:07:06
    this feels weird right so let's see um
  • 00:07:10
    so originally I used this right so I
  • 00:07:12
    would just say okay let me ask a Gemini
  • 00:07:15
    so um what does uh this stand
  • 00:07:19
    for stand for um in ASA
  • 00:07:23
    format
  • 00:07:25
    uh kind computer science term oh okay
  • 00:07:28
    the slash character okay all right so
  • 00:07:30
    let's go back here and then I'll just
  • 00:07:33
    take replace America with Los Angeles I
  • 00:07:35
    have this URL assembled and then we're
  • 00:07:37
    going to test this again okay great yeah
  • 00:07:39
    so I have this and then I return at 200
  • 00:07:42
    and that's probably exactly the same
  • 00:07:44
    kind of response uh um um drawn over
  • 00:07:48
    here right so now I'm going to save
  • 00:07:52
    this and I go back we'll say okay this
  • 00:07:55
    API actually has the weather API right
  • 00:07:57
    so I'm going to uh go back to my
  • 00:08:00
    my dog and a fox pictures and then say
  • 00:08:03
    uh I would say okay what's the weather
  • 00:08:06
    what's the weather of Seattle right uh
  • 00:08:10
    today right
  • 00:08:11
    so send a message over here and then
  • 00:08:15
    let's say um oh okay oh that's kind of
  • 00:08:18
    cool um it's November the the 13th so
  • 00:08:22
    presumably the temperature okay it's
  • 00:08:23
    still like in in U uh in Celsius right
  • 00:08:27
    not in Fahrenheit 80 to 11 degree
  • 00:08:29
    throughout the day this is a high uh
  • 00:08:32
    probability of rain especially in the
  • 00:08:33
    early morning and late evenings uh we
  • 00:08:35
    expect rainfall of this right actually
  • 00:08:37
    it's kind of raining outside right so
  • 00:08:39
    that's yeah that's cool right so uh and
  • 00:08:41
    then let's ask what's the weather of
  • 00:08:45
    Seattle what's the the weather of uh
  • 00:08:47
    Chicago right Chicago uh
  • 00:08:52
    today
  • 00:08:54
    um oh Chicago is a a little bit a um a
  • 00:08:58
    wider range since that's not by the
  • 00:09:00
    ocean right so uh significant chance of
  • 00:09:03
    rain especially in the afternoon and
  • 00:09:04
    evening right so um um since you know we
  • 00:09:07
    remember we cat this API result I can
  • 00:09:09
    ask more question right so I would say
  • 00:09:11
    uh uh is it uh going to rain R uh today
  • 00:09:17
    in Seattle and
  • 00:09:20
    Chicago right so let's say U what the
  • 00:09:22
    API whether CH gbt actually caches the
  • 00:09:26
    result right so so is it is actually
  • 00:09:28
    expected to rent today in both Seattle
  • 00:09:30
    and Chicago right so okay let's just
  • 00:09:32
    uh uh say a double check right so I I
  • 00:09:35
    would to do um I would do Seattle
  • 00:09:38
    weather uh weather
  • 00:09:41
    today and uh oh okay actually it is
  • 00:09:44
    raining right 95% of a chance of reading
  • 00:09:47
    and then if I do a
  • 00:09:48
    Chicago uh
  • 00:09:50
    weather and I get U uh 69% of ring ring
  • 00:09:55
    chance and then by 900 p.m. it's 97%
  • 00:09:58
    right so that's actually pretty good if
  • 00:10:00
    you look at a Fahrenheit over here so
  • 00:10:02
    current currently it's 11° celsus uh
  • 00:10:06
    today right so this looks a pretty I
  • 00:10:09
    don't know uh maybe pretty accurate so
  • 00:10:11
    so we'll see right so so one trick thing
  • 00:10:14
    that we did was that converting the
  • 00:10:16
    longitude and attitude uh from Seattle
  • 00:10:19
    to Chicago I'm I'm sorry from Chicago
  • 00:10:22
    I'm sorry let me put it this again from
  • 00:10:24
    the location to Chicago right so um what
  • 00:10:27
    we're going to do is um um let's say um
  • 00:10:31
    I'll just uh I'll just enable um uh
  • 00:10:35
    what's the I'll just enable this right I
  • 00:10:38
    enable debug over here and then we're
  • 00:10:40
    going to ask uh let's do this a new uh
  • 00:10:43
    new chat what's this weather Chicago
  • 00:10:45
    today right we're going to observe the
  • 00:10:48
    API out outputs uh through the debug
  • 00:10:51
    mode right so this is actually this is
  • 00:10:53
    actually my response from C chat right
  • 00:10:56
    from 5 to 12 degrees over here and let's
  • 00:10:59
    see the API call the API call here is
  • 00:11:02
    that the latitude of 41 degrees and
  • 00:11:05
    longitude is minus 87 degrees right so
  • 00:11:07
    we don't really know and let's just uh
  • 00:11:10
    uh give it a try right so I'll go to
  • 00:11:12
    Google
  • 00:11:14
    Maps and I'm going to
  • 00:11:17
    assemble uh I'm going to assemble this
  • 00:11:20
    two locations let's make this one bigger
  • 00:11:22
    for you
  • 00:11:24
    and
  • 00:11:27
    um let's say what do the is really
  • 00:11:30
    really Chicago right so Google Maps
  • 00:11:33
    promps me with this and oh where is this
  • 00:11:37
    oh it is it is in it is in Chicago right
  • 00:11:41
    so uh cad actually finds out uh oh a
  • 00:11:45
    very heart of Chicago right uh the
  • 00:11:47
    longitude and the ltitude of latitude
  • 00:11:50
    and longitude of Chicago and then
  • 00:11:52
    automatically replace this right this is
  • 00:11:54
    all due to uh this uh description right
  • 00:11:59
    this this an instruction over here right
  • 00:12:01
    so uh I said convert the location to
  • 00:12:04
    Latitude uh and uh this is the location
  • 00:12:07
    or latitude right and convert the
  • 00:12:09
    location to longitude and then we'll get
  • 00:12:10
    we we get we get all of this right so
  • 00:12:13
    great this this actually works and this
  • 00:12:15
    has been the four tutorial of using the
  • 00:12:17
    advanced mode of customer GPD tool where
  • 00:12:20
    uh you can uh assemble a complex query
  • 00:12:23
    over here and then get some WEA
  • 00:12:25
    information by using um uh cchat but
  • 00:12:29
    thank you very much for watching and
  • 00:12:30
    then I'll stay tuned uh and see you in
  • 00:12:33
    the next tutorial thank you
Etiquetas
  • GPD Tools
  • Weather API
  • Complex Queries
  • CChat
  • API Parameters
  • Latitude
  • Longitude
  • Debugging
  • Product Management
  • Advanced Tutorial