Building AI Applications with OCI Generative AI Service

OCI Generative AI Service brings enterprise-grade large language model capabilities to Oracle Cloud, with a focus on security, compliance, and integration with existing Oracle workloads.

Available Models

The service provides access to multiple foundation models. Cohere Command models excel at text generation, summarization, and classification tasks. Cohere Embed models generate high-quality text embeddings for semantic search and RAG applications. Meta Llama models offer open-weight alternatives with strong general-purpose capabilities.

All models run within your OCI tenancy boundary, ensuring data privacy and compliance with enterprise security policies.

Prompt Engineering for OCI Workloads

Effective prompts for enterprise applications follow a structured format:

from oci.generative_ai_inference import GenerativeAiInferenceClient
from oci.generative_ai_inference.models import CohereChatRequest

client = GenerativeAiInferenceClient(config=oci.config.from_file())

chat_request = CohereChatRequest(
    message="Summarize the key OCI cost optimization strategies",
    max_tokens=500,
    temperature=0.3,
    preamble="You are an Oracle Cloud expert assistant."
)

Building RAG Pipelines

Retrieval-Augmented Generation (RAG) combines the knowledge retrieval capabilities of vector databases with the generation capabilities of LLMs. OCI supports RAG architectures using OpenSearch (for vector storage), Generative AI Embeddings, and Generative AI Chat models.

Integration with Oracle Database 23ai

Oracle Database 23ai includes native vector search capabilities, enabling RAG pipelines that leverage your existing Oracle Database investments without additional infrastructure.