Spring AI Playground

What is Spring AI Playground?

Spring AI Playground is a self-hosted web UI for AI experimentation and testing, built on Spring AI for Java developers. It runs locally on port 8080 and simplifies working with large language models (LLMs), vector databases, prompt engineering, and Model Context Protocol (MCP) integrations.

How to use Spring AI Playground?

Build and run the app with ./mvnw clean install then ./mvnw spring-boot:run, then open http://localhost:8080. Configure MCP server connections through the MCP Playground interface, select connections in the chat page, and test tools with the MCP Inspector. To switch model providers, modify pom.xml and application.yaml (e.g., Ollama to OpenAI).

Key features of Spring AI Playground

• MCP Playground with connection management (STREAMABLE HTTP, STDIO, SSE)
• MCP Inspector to explore and test external tools
• Chat interface that executes MCP tools automatically
• Vector database playground with document upload and similarity search
• RAG chat using selected documents as knowledge base
• Support for all major AI model providers (Anthropic, OpenAI, etc.)

Use cases of Spring AI Playground

• Experiment with multiple LLMs and MCP tool integrations side by side
• Prototype retrieval-augmented generation (RAG) pipelines with custom documents
• Debug and refine MCP server configurations before production deployment
• Build and test Spring AI applications with contextual awareness and external capabilities

FAQ from Spring AI Playground

What platforms and tools does Spring AI Playground support?

It runs as a self-hosted web UI on any system with Java and Maven. It supports all major AI model providers (Anthropic, OpenAI, Microsoft, Amazon, Google, Ollama) and all major vector database providers (e.g., PostgreSQL/PGVector, Pinecone, Chroma). MCP transport types include STREAMABLE HTTP, STDIO, and SSE.

Does it require an API key?

No, when using Ollama as the default provider, no API key is needed. For other providers like OpenAI, you must provide an API key in application.yaml.

How do I switch between AI model providers?

Modify the pom.xml to include the desired Spring AI starter dependency and update application.yaml with the provider’s API key and profile settings. The README uses OpenAI as a step-by-step example.

What file types can I upload for RAG?

PDFs, Word documents, and PowerPoint presentations are supported. The system extracts text, chunks it, and generates embeddings for semantic search.

Are there any known limitations for MCP tool integration?

Yes, when using Ollama as the AI provider, you must use a tool-enabled model (e.g., Qwen 3 or DeepSeek-R1) that supports external function calling. Not all Ollama models work with MCP tools.