Skip to content

WMS ID #11686 - Updated screenshots and details for the new AI Agents OCI interface #394

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 11 commits into from
Apr 17, 2025
Merged
5 changes: 3 additions & 2 deletions multichannel-oci-genaiagents/2-setup/setup.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,8 @@ This lab assumes you have:

## Task 1: Ensure Chicago region subscription

The OCI Generative AI Agents service is currently only available in the Chicago region.
The OCI Generative AI Agents service is currently available in the following regions: Chicago, London, Frankfurt, Osaka and Sao Paulo.
The workshop includes prebuilt scripts compatible only with the Chicago region.
If your tenancy is already subscribed to the Chicago region, please skip to the next task.

1. On the top right, click the Regions drop down menu.
Expand Down Expand Up @@ -156,4 +157,4 @@ Next, we will create the access policies:
* **Author** - Anshuman Panda, Principal Generative AI Specialist, Alexandru Negrea
, AI and App Integration Specialist Leader

* **Last Updated By/Date** - Anshuman Panda, Principal Generative AI Specialist, Aug 2024
* **Last Updated By/Date** - Vlad Man, Digital Adoption Manager - AI Initiatives, April 2025
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,9 @@ This lab assumes you have:

1. Make sure that the root compartment is selected in the **Compartment** list.

1. Make sure that the **Object storage** option is selected in the **Select data store** list.
1. Make sure that the **Object storage** option is selected in the **Data Store type** list.

1. Click the **Enable hybrid search** option to enhance the results of the search.

1. Click the **Specify data source** button at the top of the **Data sources** table.

Expand All @@ -47,11 +49,11 @@ This lab assumes you have:

1. Select the storage bucket into which you've uploaded the dataset text files in the previous lab.

1. Select the **Select all in bucket** option from the **Object prefixes** list.
1. Select the **Select all in bucket** option.

1. Click the **Create** button at the bottom of the pane.

![Create knowledge base and data source](./images/create-data-source.png)
![Create knowledge base and data source](./images/create-data-source-updated.png)

1. Make sure that the **Automatically start ingestion job for above data sources** option is checked.

Expand Down Expand Up @@ -103,4 +105,4 @@ Here are the steps to monitor the Ingestion Job's progress.

* **Author** - Anshuman Panda, Principal Generative AI Specialist, Alexandru Negrea, AI and App Integration Specialist Leader

* **Last Updated By/Date** - Anshuman Panda, Principal Generative AI Specialist, Aug 2024
* **Last Updated By/Date** - Vlad Man, Digital Adoption Manager - AI Initiatives, April 2025
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
54 changes: 45 additions & 9 deletions multichannel-oci-genaiagents/5-create-an-agent/create-an-agent.md
Original file line number Diff line number Diff line change
Expand Up @@ -31,21 +31,57 @@ This lab assumes you have:

![Create Agent navigation](./images/create-agent-navigation.png)

1. Provide a name for the Agent (for example: oci-generative-ai-agents-cw24-hol-agent)
1. Provide a name for the Agent (for example: oci-generative-ai-agents-agent-ko4)

1. Make sure that the root compartment is selected in the **Compartment** list.

1. Optionally, provide a **Welcome message** for the Agent to display at the start of a new conversation (also called the `Preamble`, for example: Hello, i'm the OCI Generative AI documentation helper! How can i help you today?).

1. Under the **Add knowledge bases** section, make sure that the root compartment is selected in the **Compartments** list.
1. Click the **Next** button on the bottom of the page.

1. Check the box next to the knowledge base we have created in the previous lab to let the Agent know it should interact with the data specified in the knowledge base.
![Create Agent - Basic Information](./images/create-agent-basic-information.png)

1. Make sure that the **Automatically create an endpoint for this agent** option is checked.
1. On the next page click on **Create tool**.

1. Click the **Create** button at the bottom of the page.
![Create Agent - Add tool](./images/create-agent-add-tool.png)

![Create knowledge base and data source](./images/create-agent.png)
1. In the next page select **RAG** as tool type.

1. Provide a name for the tool (for example: oci-generative-ai-agents-tool-ko4) and a description. This description is important for the agent, it outlines the purpose of this tool. In case you have multiple tools, it is recommended to have a good and specific description which allows to differentiate clearly between them. For example, as "bad practice", a tool description should not be "Tool description 1". A proper name would be "Tool contains knowledge about the following topic: Leave policies".

1. In the **Add knowledge bases** section, select the right compartment and then click on the knowledge base we created earlier.

1. You can finish this step by clicking the **Create tool** button.

![Create Agent - Create tool](./images/create-agent-create-tool.png)

1. After the tool is created, proceed to the next step by clicking **Next**

![Create Agent - Confirm tool](./images/create-agent-confirm-tool.png)

1. For the **Setup agent endpoint** section we can have a lot of options for configuration.

1. First select the **Automatically create an endpoint for this agent** option.

1. For this laboratory, make sure that the **Enable human in the loop** option is **Disabled**.

1. We can proceed by configuring **Guardrails**, which are safety measures for our agent. The recommended option for each of them is the middle one, which aims to **Block** malicious intents.

1. First we select the Guardrails for **Content moderation**. You can click the options in there to select a specific level.

![Create Agent - Setup agent endpoint Content moderation Guardrails](./images/create-agent-setup-endpoint-1.png)

1. Scroll down the page and you will find **Prompt injection (PI) protection** and **Personally identifiable information (PII) protection**.

1. Select **Block** for all the options.

1. To finish this step, click the **Next** button.

![Create Agent - Setup agent endpoint PI & PII Guardrails](./images/create-agent-setup-endpoint-2.png)

1. For the final step, review the **Basic information**, **Tools** and **Endpoint configuration** and the click on the **Create agent** button.

![Create Agent - Setup agent review](./images/create-agent-review.png)

If everything went to plan, your Agent will be created. This can take a few minutes.

Expand All @@ -65,15 +101,15 @@ Please wait until the **Lifecycle state** shows the **Active** state before movi

1. Agent **name**.
2. Agent **OCID**.
3. **Knowledge Bases** associated with the Agent.
3. **Tools** associated with the Agent.
4. **Endpoints** which can be used to access the Agent programmatically (a default one was created when we created the Agent).

> In addition, you can **Launch a chat** session with the Agent as well as **Edit**, **Move** and **Delete** the Agent.

![Agent details](./images/agent-details.png)
![Agent details](./images/inspect-agent-metadata.png)

## Acknowledgements

* **Author** - Anshuman Panda, Principal Generative AI Specialist, Alexandru Negrea, AI and App Integration Specialist Leader

* **Last Updated By/Date** - Anshuman Panda, Principal Generative AI Specialist, Aug 2024
* **Last Updated By/Date** - Vlad Man, Digital Adoption Manager - AI Initiatives, April 2025
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Original file line number Diff line number Diff line change
Expand Up @@ -33,29 +33,29 @@ This lab assumes you have:

1. You should see the greeting message we've entered during the Agent creation displayed in the chat section.

![Start chatting](./images/start-chat.png)
![Start chatting](./images/start-chat-updated.png)

1. At this point, we can start asking our Agent questions about the data we provided and get intelligent answers as well as references to the original data that the answers are based on.

Let's type the following question into the chat: "How can i create a fine-tuned model?" and click the **Submit** button.
Let's type the following question into the chat: "How many free days do i have?" and click the **Submit** button.

![Ask first question](./images/ask-first-question.png)
![Ask first question](./images/ask-first-question-updated.png)

1. The Agent will scan the data for relevant information and compose a similar to the following:

![First answer](./images/first-answer.png)
![First answer](./images/first-answer-updated.png)

1. The Agent can also provide a direct reference to the data in our dataset where the answer was extracted from.

Scroll down to the end of the answer text and click **View citations** to expand the citations section. This section will provide one or more references to our text files which include a link to the file and an excerpt from the file where the answer was extracted from.

Providing citations makes sure that the Agent bases it's responses on our data and decreases the chances for hallucinations or made up answers.

![First answer citations](./images/first-answer-citations.png)
![First answer citations](./images/first-answer-citations-updated.png)

In addition to citations, you can also observe the log section on the right of the screen to which search query the Agent is using as well as which data files were found to have relevant answers and the text generated for the response.
In addition to citations, you can also observe the **Traces** section on the right of the screen to see the internal thinking process of the agent, planning, which search query the Agent is using as well as which data files were found to have relevant answers and the text generated for the response.

![Logs](./images/logs.png)
![Traces](./images/traces.png)

Feel free to experiment and ask the Agent additional questions related to your uploaded document.

Expand All @@ -64,4 +64,4 @@ Feel free to experiment and ask the Agent additional questions related to your u

* **Author** - Anshuman Panda, Principal Generative AI Specialist, Alexandru Negrea, AI and App Integration Specialist Leader

* **Last Updated By/Date** - Anshuman Panda, Principal Generative AI Specialist, Aug 2024
* **Last Updated By/Date** - Vlad Man, Digital Adoption Manager - AI Initiatives, April 2025
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.