AI Optimizer Client

The included web-based client is built with Streamlit and communicates with the AI Optimizer Server through its REST API. It is a workbench for configuring and exercising the AI Optimizer's agents, flows, and MCP-backed tools.

BYO Client

You can replace the included client with any client that can communicate with the REST API.

The client provides a practical interface for working with the AI Optimizer. You can:

• Run LLM-only and tool-enabled agent routes
• Evaluate agent and flow responses
• Experiment with prompt engineering
• Split and embed documents
• Configure Deep Data Security
• Inspect portable AgentSpec definitions
• Set up databases, models, and access to Oracle Cloud Infrastructure

Choose a Path

• Run an LLM-only Agent: Configure a language model, then open the Chatbot.
• Run an NL2SQL Agent: Configure an Oracle AI Database, make SQLcl MCP Server available, then enable NL2SQL in the Chatbot.
• Run a grounded Vector Search Flow: Configure an Oracle AI Database and language and embedding models. Then use Split/Embed to create a vector store.
• Evaluate agent routes with data: Use the Testbed to compare model, prompt, and tool configurations against your data.

The Chatbot's Tool Selection controls which route handles each request. See Agents and Flows for the runtime details.

Error alerts show the exception name above the provider's message. Repeated exception prefixes and masked API-key values are removed for readability; the provider's wording and links are preserved.