Getting help from the AI assistant

IBM Cloud®'s AI assistant, which is powered by IBM®'s watsonx, is designed to help you learn about working in IBM Cloud and building solutions with the available catalog of offerings.

The AI assistant is designed as a retrieval-augmented generation (RAG) implementation that is running against IBM® large language models (LLMs) using IBM's watsonx and the IBM Cloud docs.

IBM Cloud does not gather any personal identifiable information when you are using the AI feature and it is not connected to or associated with your account. When you query the AI assistant, it is not being trained based on user input and does not associate any queries with your identity or account. IBM endeavors to develop AI in accordance with its published principles of trust and transparency. For more information, see AI Ethics.

The AI assistant supports all national languages supported by IBM Cloud including Brazilian Portuguese, English, French, German, Japanese, Korean, Italian, Spanish, Simplified Chinese, and Traditional Chinese. To get responses from the AI assistant in one of the supported languages, ask your question in one of the languages and get an answer back in that same language.

Using the AI assistant in the IBM Cloud console

No matter where you are working in the IBM Cloud console, the AI assistant is available to help with your questions about IBM Cloud so that you can learn more about the platform or the services that you're using.

You can start the AI assistant from the IBM Cloud console header from the Help menu anytime that you want to ask a question, and it follows along with you in the console until you close it. With each question that you ask, reference citations from the IBM Cloud docs are provided.

From the IBM Cloud console menu bar, click the Help icon > AI assistant.
Ask questions about working in your account, working with products and offerings from the catalog, and more to help you stay productive in IBM Cloud.
Use the supported documentation references that are returned to have access to additional information and the ability to verify responses.

Content that is generated in the AI assistant might include mistakes or be incorrect.

Submitting feedback for generated responses

You can provide feedback for each response from the AI assistant by using the Good response icon Good response icon and Bad response icon Bad response icon options. You can choose to just mark the provided answer as a good response or bad response, and you can provide additional feedback by selecting from a common set of qualities that describe why you chose the rating and add additional feedback in the form of a comment. Your feedback is not used to train or enhance AI models.

Clearing your chat history

All of the questions that you ask and the generated responses are retained in the AI assistant until you decide that you want to clear the history. To clear the history, open the AI assistant, and then click Open menu icon Open menu icon > Clear history.

Closing the chat

The AI assistant follows along with you in the console until you close it. You can move it to position it wherever you'd like on the page. When you're ready to close it, click the Close icon . Your history is saved even when you close the AI assistant, so you can see your previous questions and answers the next time that you open it.

Using the AI assistant in the IBM Cloud CLI

You can also ask questions in the IBM Cloud CLI.

Log in with the ibmcloud login command. If you are logging in with a federated ID, use the --sso option to authenticate with a one-time passcode, or use the --apikey option to authenticate with an API key.
Ask a question with the ibmcloud assist command.
```
ibmcloud assist "How do I update the CLI?"
```

For more information about using the AI assistant in the IBM Cloud CLI, see General IBM Cloud CLI (ibmcloud) commands.

Limitations

The AI assistant has the following limitations:

Questions must be limited to 500 characters or less.
AI-generated responses are limited to information from the IBM Cloud docs and the LLM.
Actions like creating resources or making changes within your account can't be completed by the AI assistant.
Answers from the previous query and response are not used to answer subsequent queries.
If you encounter any styling issues in the console, use the Open menu icon > Clear history option, and refresh the page.

Tips for creating effective prompts

When you ask the AI assistant a question, you are creating a prompt that is used to generate an answer. To ensure that you get the best response, review the following tips:

Always include context: The more specific that you can be by including the right details, the better the response is. For example, if you're asking about a dashboard within the console for a specific service, you want to make sure to include the name of that service.
Ask questions that can be answered by the AI assistant: The AI assistant is limited to answering questions about IBM Cloud. This means that your questions should remain about IBM Cloud and its offerings.
Be clear and concise: You want to include enough detail to get your question answered correctly. But including too much detail or being too verbose in your prompt might result in an incorrect answer.
Be kind and polite: Everyone wants to be treated kindly, even AI assistants.
Rephrase your question when needed: If you don't get the answer that you expected, sometimes you need to rework your prompt to provide more or less detail or use different terms and synonyms to get the right response. For example, if you are looking for CLI instructions instead of console instructions, you might need to specify that you want information about a specific command to complete your task.

Here are some examples of questions that you might ask:

What's the difference between Pay-As-You-Go and Subscription accounts?
How can I increase my quotas for VPC resources?
How should I organize my secrets?
Help me learn more about Power Virtual Server