OCI Generative AI Overview

Use Enterprise AI Models when you want model access for inference tasks such as conversational generation, semantic search, recommendation, classification, clustering, and relevance ranking.

OCI Generative AI supports:

• Chat for conversational experiences such as question answering and virtual assistants
• Embeddings for semantic search, recommendation, classification, and clustering
• Rerank for ordering documents by relevance to a query
• OpenAI-compatible APIs for integrating with existing tools and SDKs

You can use OCI Generative AI models in several ways. You can call pretrained hosted models through the Console, API, or CLI. You can also import, fine-tune, and host custom models on dedicated AI clusters. This gives you a path from experimentation to production with enterprise controls and deployment flexibility.

Use Enterprise AI Agents when you want to build production-grade agentic applications that combine models with tools, memory, retrieval, and orchestration.

OCI Generative AI provides two main approaches for building agents:

• Build agents with the OCI Responses API
• Deploy hosted agentic applications in OCI Generative AI

You can also combine these approaches in a hybrid architecture.

The OCI Responses API is the primary API for agentic workflows. It is OpenAI-compatible and supports model interaction, orchestration, reasoning, conversation state, and tool use. Supported tools include File Search, Code Interpreter, Function Calling, and MCP Calling. Agents can also use supporting resources such as Files, Vector Stores, Containers, Conversations, Projects, and memory features such as long-term memory and short-term memory compaction.

OCI Generative AI also supports SQL Search (NL2SQL) for agent workflows that need structured enterprise data access. NL2SQL converts natural-language requests into validated SQL by using semantic enrichment and structured data metadata.

For teams that want OCI-managed hosting for custom runtimes, OCI Generative AI provides Applications and Deployments for hosted agentic applications. This model supports container-based deployment, managed infrastructure, networking, storage integration, and identity configuration.

Use Enterprise AI Governance when you need to secure and control how generative AI resources are accessed, deployed, and used.

Enterprise AI governance in OCI Generative AI combines infrastructure, identity, network security, and runtime controls to help keep AI systems secure, compliant, and aligned with organizational policy.

Key governance capabilities include:

• IAM Policies for controlling who can access, use, and manage OCI Generative AI resources
• Private Endpoints for keeping model access within a secure network boundary
• API Keys for accessing OCI Generative AI services
• OAuth for agentic tasks that require OCI IAM identity-domain application integration
• Zero Trust Packet Routing (ZPR) for identity-based network enforcement
• Guardrails for applying runtime safety and compliance controls to model inputs and outputs

Together, these capabilities provide end-to-end governance across access control, network security, hosted application configuration, and AI behavior.

These three areas work together as part of one platform:

• Enterprise AI Models provide the model foundation for inference and generation
• Enterprise AI Agents add orchestration, tools, memory, retrieval, and hosted execution
• Enterprise AI Governance applies the security, access, and compliance controls needed for enterprise use

This combination lets you move from model access, to agentic application development, to production deployment with governance built in.

• Enterprise AI Models in OCI Generative AI
• Enterprise AI Agents in OCI Generative AI
• Enterprise AI Governance for OCI Generative AI