Creating RAG chat bot with Azure AI Studio
MH

Marko Himonov

Created Aug 19, 2024

Creating RAG chat bot with Azure AI Studio

105
    p

Go to Microsoft Azure portal

32 Steps
1

Type "azure ai"

Type "azure ai"
2

Click on Azure AI services

Click on Azure AI services
3

Click on Create (on the Overview tab)

Click on Create (on the Overview tab)
4

Select the group you need

Select the group you need
5

If you don't have a group resource create one

If you don't have a group resource create one
6

Write a name for your resources

Write a name for your resources
7

Click on Pricing tier

Click on Pricing tier
8

Click on Next

Click on Next
9

Click on All networks, including the internet, can access this resource.

Click on All networks, including the internet, can access this resource.
10

Click on Next

Click on Next
11

Optionally specify tags. If you don't care about tags, you can skip this step without consequences

Optionally specify tags. If you don't care about tags, you can skip this step without consequences
12

The last tab is to make sure you have entered everything correctly. Once you are sure, click create

The last tab is to make sure you have entered everything correctly. Once you are sure, click create
13

After you see "Your deployment is complete" message click on Go to resource

After you see "Your deployment is complete" message click on Go to resource
14

Click on Model deployments

Click on Model deployments
15

Click on Manage Deployments

Click on Manage Deployments
16

In the window that opens, click on Create new deployment

In the window that opens, click on Create new deployment
17

Write a name for the deployment

Write a name for the deployment
18

Click on the dropdown

Click on the dropdown
19

Select the desired model, e.g. gpt4o

Select the desired model, e.g. gpt4o
20

After you select a model, the rest of the fields on the page will be filled in by default, but you can still choose other options. Click on Create

After you select a model, the rest of the fields on the page will be filled in by default, but you can still choose other options. Click on Create
21

Click on Explore Azure AI Studio…

Click on Explore Azure AI Studio…
22

Click on Your current Azure AI Studio experience is running on an Azure OpenAI resource. To unlock all capabilities, create a hub and project.

Click on Your current Azure AI Studio experience is running on an Azure OpenAI resource. To unlock all capabilities, create a hub and project.
23

Click on Create now

Click on Create now
24

Click on Create project and continue

Click on Create project and continue
25

Click on Create a new project…

Click on Create a new project…
26

Write a name for your project

Write a name for your project
27

Click on Create a new hub

Click on Create a new hub
28

Write a name for your hub

Write a name for your hub
29

Click on dropdown to select resource group you want to assign

Click on dropdown to select resource group you want to assign
30

If you need to create new one - use the Create new resource group link

If you need to create new one - use the Create new resource group link
31

Other fields are optional and can be left unchanged. Click on Next

Other fields are optional and can be left unchanged. Click on Next
32

Click on Create a project. You need to wait and check that everything you need has been created

Click on Create a project. You need to wait and check that everything you need has been created
    p

Go to Microsoft Azure portal

32 Steps
33

Type "azure ai search" фnd select it from the drop-down list.

Type "azure ai search" фnd select it from the drop-down list.
34

Click on Create

Click on Create
35

Select the group you need

Select the group you need
36

If there is no such group, click on Create new right below the field

If there is no such group, click on Create new right below the field
37

Write a name for your project

Write a name for your project
38

Click on Change Pricing Tier

Click on Change Pricing Tier
39

Click on Pricing tiers (for free tier - maximum 50 mb of storage, 25 mb of storage for vector indexes, maximum 3 indexes, it's about 40-50 pdf files, depending on the size)

Click on Pricing tiers (for free tier - maximum 50 mb of storage, 25 mb of storage for vector indexes, maximum 3 indexes, it's about 40-50 pdf files, depending on the size)
40

Click on Select

Click on Select
41

Click on Review + create

Click on Review + create
42

Click on Settings

Click on Settings
43

Click on Add new connection

Click on Add new connection
44

Click on Azure AI Search…

Click on Azure AI Search…
45

You will see a new window, in it find the previously created Azure Ai Search

You will see a new window, in it find the previously created Azure Ai Search
46

Click on Add connection

Click on Add connection
47

Click on Close

Click on Close
48

Click on Chat

Click on Chat
49

Click on Add your data

Click on Add your data
50

Click on Add a new data source

Click on Add a new data source
51

Click on Data source

Click on Data source
52

Click on Upload files

Click on Upload files
53

Click on Upload

Click on Upload
54

Click on Upload files

Click on Upload files
55

Click on Next

Click on Next
56

Click on Select available project index

Click on Select available project index
57

Select the previously created Azure Ai Search project

Select the previously created Azure Ai Search project
58

Write a name for your index

Write a name for your index
59

Click on Next

Click on Next
60

Click on Select an embedding model. Optionally add vector search for more precise information retrieval (will create an embedding model if not already created, there is a charge for this)

Click on Select an embedding model. Optionally add vector search for more precise information retrieval (will create an embedding model if not already created, there is a charge for this)
61

Click on Create

Click on Create
62

Click on Next

Click on Next
63

Type your questions

Type your questions
64

See the result

See the result
    a

Optionally: automatic index update. Go to Azure AI Studio

17 Steps
65

Click on Add a new data source

Click on Add a new data source
66

Click on Data source

Click on Data source
67

Click on Azure Blob Storage

Click on Azure Blob Storage
68

Click on Blob store

Click on Blob store
69

Click on workspaceblobstore

Click on workspaceblobstore
70

Click on UI/

Click on UI/
71

Click on the folder that Azure Ai Studio created for us when we uploaded the files for the first time

Click on the folder that Azure Ai Studio created for us when we uploaded the files for the first time
72

Click on Next

Click on Next
73

Click on Select available project index

Click on Select available project index
74

Select your Azure Ai Search service

Select your Azure Ai Search service
75

Write name for your index

Write name for your index
76

Click on Schedule updates

Click on Schedule updates
77

Click on Schedule updates

Click on Schedule updates
78

Click on Next

Click on Next
79

Click on Daily

Click on Daily
80

Click on Create

Click on Create
81

Select Select from all options

Select Select from all options

Your index is now automatically updated every hour. You can add new documents and Azure will update the index itself

    a

Optionally: deploy your chat as a web application. Go to Azure AI Studio

16 Steps
82

Select the virtual machine you need

Select the virtual machine you need
83

Double click on cost to sort the price from low to high

Double click on cost to sort the price from low to high
84

Click on Deploy

Click on Deploy
85

Click on ...as a web app

Click on ...as a web app
86

Type the name of your web app

Type the name of your web app

The cheapest virtual machine costs $0.25 per hour. It will cost from $2 to $10 per month depending on the number and content of your pdfs

87

Click on Select a subscription and select your subscription

Click on Select a subscription and select your subscription
88

Click on dropdown to select resource group you want to assign

Click on dropdown to select resource group you want to assign
89

Click on Select a location

Click on Select a location
90

Choose a location, e.g. East US

Choose a location, e.g. East US
91

Click on Select a pricing plan

Click on Select a pricing plan
92

Select the pricing plang, e.g. Basic (B1)

Select the pricing plang, e.g. Basic (B1)
93

Optionally you can enable chat history, but this will create a CosmosDB, which will incur an additional fee

Optionally you can enable chat history, but this will create a CosmosDB, which will incur an additional fee

Your web app will be configured with Azure Active Directory authentication enabled. It may take a few minutes to apply after deployment completes, during which time you will not be able to chat in the app. Please wait 10 minutes, then reload the app and log in to begin chatting.

94

Сlick on Deploy

Сlick on Deploy
95

After you have waited for the deployment click on Launch

After you have waited for the deployment click on Launch
96

After logging into your account you can ask your questions

After logging into your account you can ask your questions
97

See the result

See the result
Well done!
Create how-to guides like this in a snap. Get Tango now.