6 Understanding AI Profiles in TSAM Plus Operations Assistant
Multiple AI Profiles will be created during deployment and can be updated later. Each AI profile is used for a specific type of task.
For each AI profile, you can configure AI Provider,LLM model,token limit and other AI provider specific settings in TSAMSetup.cfg. Parameter names are formed by combining the profile’s Parameter Prefix and the parameter suffix, for example, the LLM model for Main Agent is configured with
AI_MAIN_PROVIDER.
| Agent/Worker | Description | Parameter Prefix |
|---|---|---|
| Main Agent | Coordinates requests and routes them to the appropriate specialized agent or worker | AI_MAIN_ |
| Metrics Query | Answers questions by querying TSAM Plus metrics. | AI_METRICS_ |
| User Documentation | Answers general questions about the TSAM Plus. | AI_USERDOC_ |
| Console Documentation | Answers questions about how to use the TSAM Plus Console. | AI_CONSOLEDOC_ |
| Follow-up Questions | Generates suggested follow-up questions based on the current conversation. | AI_FOLLOWUPDOC_ |
Configuration for Main Agent is mandatory.
Table 6-1 TSAMSetup.cfg Parameters
| Parameter | Required / Conditional | Description |
|---|---|---|
AI_OCI_CRED_USER |
Required when any selected AI provider is oci. | For AI_OCI_* parameters, obtain the information from the OCI console or from your OCI tenancy administrators. |
AI_OCI_CRED_TENANCY |
Required when any selected AI provider is oci. | OCI tenancy OCID. |
AI_OCI_CRED_PRIVATEKEY |
Required when any selected AI provider is oci. | Path to an existing readable OCI API private key file. |
AI_OCI_CRED_FINGERPRINT |
Required when any selected AI provider is oci. | Fingerprint for the OCI API key. |
AI_OCI_COMPARTMENT_ID |
Required when any selected AI provider is oci. | OCI compartment OCID used by OCI Generative AI. |
AI_VECTORINDEX_SOURCE_MODE |
Required when TSAM_ENABLE_AI=Y. | Document source mode for vector index input, the value can be CLOUD or LOCAL.
CLOUD: the operations assistant uses the product-owned built-in documentation source. This is the default mode. LOCAL: the operations assistant uses Oracle DIRECTORY-backed sources on the database host. LOCAL is supported only when DB_IS_ADB=N. |
AI_MAIN_PROVIDER |
Required when TSAM_ENABLE_AI=Y | LLM provider for the main TSAM AI agent. |
AI_MAIN_MODEL |
Required when TSAM_ENABLE_AI=Y | Main Agent LLM model. Should be reasoning-capable because it handles the most complicated tasks. |
AI_MAIN_REGION |
Optional | OCI region for the Main Agent LLM deployment. |
AI_MAIN_MAX_TOKENS |
Optional | Maximum token limit configured for the Main Agent LLM. |
AI_METRICS_PROVIDER |
Optional | LLM provider for Metrics Query functionality. |
AI_METRICS_MODEL |
Optional | LLM model used for Metrics Query functionality. |
AI_METRICS_REGION |
Optional | OCI region for the Metrics Query LLM deployment. |
AI_METRICS_MAX_TOKENS |
Optional | Maximum token limit configured for the Metrics Query LLM deployment |
AI_USERDOC_PROVIDER |
Optional | LLM provider for User Documentation. |
AI_USERDOC_MODEL |
Optional | LLM model for User Documentation. |
AI_USERDOC_REGION |
Optional | OCI region for User Documentation. |
AI_USERDOC_MAX_TOKENS |
Optional | Maximum token limit configured for the user document QA LLM deployment User Documentation. |
AI_CONSOLEDOC_MODEL |
Optional | LLM model used for Console Documentation. |
AI_CONSOLEDOC_PROVIDER |
Optional | LLM provider for Console Documentation. |
AI_CONSOLEDOC_REGION |
Optional | OCI region for Console Documentation. |
AI_CONSOLEDOC_MAX_TOKENS |
Optional | Maximum token limit configured for Console Documentation. |
AI_FOLLOWUPDOC_PROVIDER |
Optional | LLM provider for Follow-up Questions. |
AI_FOLLOWUPDOC_MODEL |
Optional | LLM model for Follow-up Questions. |
AI_FOLLOWUPDOC_REGION |
Optional | OCI region for Follow-up Questions. |
AI_FOLLOWUPDOC_MAX_TOKENS |
Optional | Maximum token limit configured for Follow-up Questions. |
AI_OPENAI_CREDENTIAL_USERNAME |
Required when any selected AI provider is openai | OpenAI credential username. |
AI_OPENAI_API_KEY |
Required when any selected AI provider is openai | OpenAI API key. |
AI_AZURE_CREDENTIAL_USERNAME |
Required when any selected AI provider is azure. | Azure credential username. |
AI_AZURE_API_KEY |
Required when any selected AI provider is azure. | Azure API key. |
AI_AZURE_RESOURCE_NAME |
Required when any selected AI provider is azure | Azure resource name. |
AI_AZURE_EMBEDDING_DEPLOYMENT_NAME |
Required when any selected AI provider is azure | Optional Azure embedding deployment name. It is rendered only when explicitly set. |
Configuration of the other AI profiles is optional. If not explicitly configured, the following rules apply:
- PROVIDER defaults to the value of AI_MAIN_PROVIDER.
- If the AI profile uses the same AI provider with Main Agent, it also inherits the Main Agent’s LLM model and other settings.
- If the AI profile uses a different AI provider from Main Agent, you must explicitly configure its model and other provider-specific settings. Specifically, The AI_MAX_TOKENS setting is independent for each profile, and does not inherit from AI_MAIN_MAX_TOKENS, if not set, the default value is 1024.