Use when sending files (PDFs, images, documents) to an AI model from Oracle PL/SQL with UC AI — multimodal analysis via uc_ai.generate_text with a manually built message array, using uc_ai_message_api.create_file_content (BLOB or base64 CLOB),…
Use when calling an AI/LLM (OpenAI, Anthropic Claude, Google Gemini, Ollama, OCI, xAI, Mistral, OpenRouter) from Oracle PL/SQL with the UC AI library — first uc_ai.generate_text call, choosing provider/model constants, setting up API keys (uc_ai_get_key or…
Use when enabling reasoning / extended thinking for AI calls from Oracle PL/SQL with UC AI — uc_ai.g_enable_reasoning and uc_ai.g_reasoning_level (c_reasoning_level_low/medium/high), provider tuning via uc_ai_openai.g_reasoning_effort,…
Use when an AI/LLM response must be machine-parseable JSON from Oracle PL/SQL with UC AI — pass p_response_json_schema to uc_ai.generate_text to enforce a JSON Schema (classification with enums, data extraction, confidence scoring), author draft-07 schemas,…
Use when giving an AI/LLM access to database data or actions (tools / function calling) from Oracle PL/SQL with UC AI — registering tools with uc_ai_tools_api.create_tool_from_schema or merge_tool_from_schema, describing parameters with a JSON schema,…
Use when an AI agent must remember something across conversations in Oracle PL/SQL with UC AI — enabling the MEMORY tool with uc_ai_memory.enable_for_agent, choosing a store scope (agent, user, session, shared, global, context), one memory per document with a…
Use when building multi-agent AI systems in Oracle PL/SQL with UC AI — creating agents with uc_ai_agents_api.create_agent, running them via execute_agent with session IDs, sequential/loop/conditional workflow definitions, orchestrator agents that delegate to…
Use when managing reusable, versioned AI prompt templates in Oracle PL/SQL with UC AI — uc_ai_prompt_profiles_api.create_prompt_profile, execute_profile with {placeholder} parameters, model_config_json settings, runtime overrides (p_provider_override,…