Use AI Keyword to Enter Prompts

Use AI as the keyword in a SELECT statement for interacting with the database using natural language prompts.

The AI keyword in a SELECT statement instructs the SQL execution engine to use the LLM identified in the active AI profile to process natural language and to generate SQL.

You can use the AI keyword in a query with Oracle clients such as SQL Developer, OML Notebooks, and third-party tools, to interact with database in natural language.

Note: You cannot run PL/SQL statements, DDL statements, or DML statements using the AI keyword.

Syntax

The syntax for running AI prompt is:

SELECT AI `action` `natural_language_prompt`

Parameters

The following are the parameters available for the action parameter:

Parameter	Description
`runsql`	Runs the underlying SQL command for the natural language prompt and supports retrieval-augmented generation (RAG). This is the default action and does not require specifying this parameter.
`showsql`	Displays the SQL statement for a natural language prompt.
`explainsql`	Explains the generated SQL from the prompt in a natural language. This option sends the generated SQL to the AI provider to produce a natural language explanation.
`narrate`	The `narrate` action supports both natural language to SQL (NL2SQL) and RAG. For NL2SQL, `narrate` sends the result of a SQL query run by the database to the LLM, which generates a natural language description of that result. For RAG, when the AI profile includes a vector index, the system uses the specified transformer (or default transformer) model to create a vector embedding from the prompt for semantic similarity search against the vector store. The system then adds the retrieved content from the vector store to the user prompt and sends it to the LLM to generate a response based on this information. If you do not want table data or vector search documents to be sent to an LLM, a user with administrator privileges can disable such access for all users of the given database. This, in effect, disables the `narrate` action.
`chat`	Passes the user prompt directly to the LLM to generate a response, which is provided to the user. For session-based short-term conversations, if `conversation` in the `DBMS_CLOUD_AI.CREATE_PROFILE` function is set to `true`, this option includes content from prior interactions or prompts, potentially including schema metadata. For multiple named conversations, only the chat-related prompt histories are sent to the LLM. See Select AI Conversations to learn more..
`showprompt`	This action supports NL2SQL and RAG. This action does not support synthetic data generation, `explainsql`, and `narrate`. Display the constructed prompt that would be sent to the generative AI model.
`summarize`	Generates a summary of the content for your text and large files using user-specified AI providers. If your AI profile is set in the session then, you can use `SELECT AI SUMMARIZE` . See Generate a Summary with Select AI and Example: Select AI Actions to learn more. If you want to customize the summary generation, use the `DBMS_CLOUD_AI.SUMMARIZE` function. See Example: Select AI Summarize to learn more.
`feedback`	Enables you to provide feedback in natural language to improve the LLM's ability to generate more accurate SQL queries. You can use it alongside with the other Select AI's actions such as `runsql`, `showsql`, and `explainsql`. See Feedback to learn more. Note: This feature is available only on Oracle AI Database 26ai. If you are using `narrate`, ensure that the profile is configured for SQL generation and not RAG.
`translate`	Enables you to use generative AI from the OCI translation service to translate your prompt into the preferred language specified as the `target_language` attribute in your AI profile.
`agent`	Calls the agent team to process the natural language prompt. See Select AI Agent and Examples of Using Select AI Agent for more details.

To learn how you can use these actions, see Examples of Using Select AI.

Usage Notes

The AI keyword for Select AI is not supported in Database Actions or APEX Service. You can use only theDBMS_CLOUD_AI.GENERATE function.

Do not use DBMS_CLOUD_AI.SET_PROFILE in Database Actions or APEX Service. Instead, set the AI profile in the profile_name argument of DBMS_CLOUD_AI.GENERATE.
The AI keyword is supported only in a SELECT statement.
You cannot run PL/SQL statements, DDL statements, or DML statements using the AI keyword.
The sequence is SELECT followed by AI. These keywords are not case-sensitive. After setting your AI profile using DBMS_CLOUD_AI.SET_PROFILE, and optional action, the text after SELECT AI is a natural language prompt. If an AI profile is not set, SELECT AI reports the following error when running a SELECT AI statement:
```
ORA-00923: FROM keyword not found where expected
      00923. 00000 -  "FROM keyword not found where expected"
```
Special character usage rules apply according to Oracle guidelines. For example, use single quotes twice if you are using an apostrophe in a sentence.
```
select ai how many customers in SF don''t own their own home
```
LLMs are subject to hallucinations and results are not always correct:
- It is possible that SELECT AI may not be able to run the generated SQL query for a specific natural language prompt.
- It is possible that SELECT AI may not be able to generate a SQL query for a specific natural language prompt.
- It is possible that the LLM may not generate a SQL query that produces a correct result given your natural language prompt.
In such a scenario, SELECT AI may respond with information to assist you in generating valid SQL.
For better results with natural language to SQL generation, use database views or tables with contextual column names or consider adding column comments explaining values stored in the columns.
Use the explainsql action, as in SELECT AI explainsql, to get a more detailed explanation of the SQL statement, as compared to SELECT AI showsql.
To access DBA or USER views, See DBMS_CLOUD_AI Views.