- Step1: Sign up for a Magma Deploy account and verify your email.
- Step2: Create a new agent project and choose a template or start from scratch.
- Step3: Connect your data sources (documents, databases, APIs) and vector store.
- Step4: Drag and drop workflow blocks to define embeddings, retrieval, prompts, and model routing.
- Step5: Select LLM providers (OpenAI, Anthropic, or self-hosted) and configure API keys.
- Step6: Test your agent in the built-in chat emulator and adjust prompts or workflows.
- Step7: Deploy the agent as a REST endpoint or embed widget into your web app.
- Step8: Monitor usage metrics and logs in the dashboard, then iterate on workflows.