|
| 1 | +# 🚀 OCI Enterprise AI Agents |
| 2 | + |
| 3 | +Build, orchestrate, and run **production-grade AI agents** on Oracle Cloud Infrastructure (OCI). |
| 4 | + |
| 5 | +OCI Enterprise AI Agents (formerly *Agent Hub*) is Oracle’s fully managed platform for designing, deploying, and operating **enterprise-ready, tool-using, multi-step AI agents** with built-in orchestration, memory, governance, and observability. |
| 6 | + |
| 7 | +--- |
| 8 | + |
| 9 | +## 📌 What are Enterprise AI Agents? |
| 10 | + |
| 11 | +Enterprise AI Agents are **goal-driven systems** that can: |
| 12 | +- Reason through complex tasks |
| 13 | +- Use tools (RAG, search, APIs, code execution) |
| 14 | +- Maintain memory across interactions |
| 15 | +- Execute multi-step workflows |
| 16 | +- Deliver real business outcomes (not just responses) |
| 17 | + |
| 18 | +Unlike chatbots, these agents **take actions** and integrate deeply with enterprise systems. |
| 19 | + |
| 20 | +--- |
| 21 | + |
| 22 | +## ✨ Key Capabilities |
| 23 | + |
| 24 | +### 🧠 Agent Runtime |
| 25 | +- Multi-step reasoning & orchestration |
| 26 | +- Tool invocation (RAG, NL2SQL, APIs, Code Interpreter) |
| 27 | +- Stateful execution with short-term & long-term memory |
| 28 | + |
| 29 | +### 🔌 Built-in Tools |
| 30 | +- Web Search |
| 31 | +- File Search (RAG) |
| 32 | +- NL2SQL |
| 33 | +- Code Interpreter |
| 34 | +- Function Calling / API integration |
| 35 | +- MCP (Model Context Protocol) support |
| 36 | + |
| 37 | +### 📦 Managed Infrastructure |
| 38 | +- Vector Stores (fully managed) |
| 39 | +- File & data handling APIs |
| 40 | +- Secure execution environment |
| 41 | + |
| 42 | +### 🔐 Enterprise-Grade Governance |
| 43 | +- OCI IAM integration |
| 44 | +- Audit logging & traceability |
| 45 | +- Policy enforcement & guardrails |
| 46 | + |
| 47 | +### 📊 Observability |
| 48 | +- OpenTelemetry-based tracing |
| 49 | +- (Upcoming) Managed Langfuse integration |
| 50 | + |
| 51 | +--- |
| 52 | + |
| 53 | +## 🧑💻 Who is this for? |
| 54 | + |
| 55 | +- Developers & Architects building agentic workflows |
| 56 | +- Enterprise IT teams automating operations |
| 57 | +- Business teams (HR, Finance, CX, Supply Chain) |
| 58 | +- AI/ML teams building production-grade GenAI systems |
| 59 | + |
| 60 | +--- |
| 61 | + |
| 62 | +## 🛠️ Getting Started |
| 63 | + |
| 64 | +### 🔗 Useful Examples |
| 65 | + |
| 66 | +Explore official Oracle samples for working with OCI Enterprise AI Agents (authentication, SDK usage, and model calls): |
| 67 | + |
| 68 | +👉 https://github.com/oracle-samples/oci-genai-auth-python/tree/main/examples |
| 69 | +👉 https://docs.oracle.com/en-us/iaas/Content/generative-ai/agents.htm |
| 70 | + |
| 71 | +### Prerequisites |
| 72 | +- OCI Account |
| 73 | +- Access to OCI Generative AI / Enterprise AI services |
| 74 | +- Python 3.9+ (or your preferred language) |
| 75 | + |
| 76 | +--- |
| 77 | + |
| 78 | +## 🔄 Deployment Options |
| 79 | + |
| 80 | +OCI Enterprise AI Agents supports flexible deployment: |
| 81 | + |
| 82 | +- Self-Hosted – Full control within your environment |
| 83 | +- Customer-Managed – OCI-managed setup with your control |
| 84 | +- Fully Hosted – Fully managed by Oracle |
| 85 | + |
| 86 | +--- |
| 87 | + |
| 88 | +## 🌍 Open & Flexible |
| 89 | + |
| 90 | +Supports open source frameworks like: |
| 91 | +- OpenAI Agents SDK |
| 92 | +- OpenAI Codex SDK |
| 93 | +- Microsoft Agent Framework |
| 94 | + -LlamaIndex |
| 95 | + -Pydantic |
| 96 | + -CrewAI |
| 97 | +- LangChain |
| 98 | +- LangGraph |
| 99 | +- AutoGen |
| 100 | +- LlamaIndex |
| 101 | + |
| 102 | +Supports open standards: |
| 103 | +- MCP (Model Context Protocol) |
| 104 | +- A2A (Agent-to-Agent) |
| 105 | + |
| 106 | +--- |
| 107 | + |
| 108 | +## 📈 Roadmap (Highlights) |
| 109 | + |
| 110 | +- Managed Langfuse (Observability) |
| 111 | +- GenAI Skills API |
| 112 | +- Agent Builder (Low-code / No-code) |
| 113 | +- Dynamic Model Routing |
| 114 | +- GenAI Sandboxes |
| 115 | + |
| 116 | +--- |
| 117 | + |
| 118 | +## 📄 License |
| 119 | + |
| 120 | +MIT License (or your preferred license) |
0 commit comments