plan_multimodal_retrieval
Read-only
Plan a multimodal retrieval rollout for visual and document artifacts, configuring goal, evidence types, corpus size, embedding dimensions, latency budget, and reranker usage to avoid GPU training.
Instructions
Plan a high-ROI multimodal retrieval rollout for screenshots, PDF pages, dashboard captures, and proof artifacts without starting GPU training.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| goal | No | Business or workflow objective for visual/document retrieval. | |
| evidenceTypes | No | Evidence surfaces to include, such as screenshots, pdf_pages, proof_artifacts, dashboards, or videos. | |
| corpusItems | No | Estimated number of visual artifacts or document pages to index. | |
| maxEmbeddingDim | No | Maximum embedding dimension to budget for Matryoshka-style truncation planning. | |
| latencyBudgetMs | No | Target retrieval latency budget for agent recall. | |
| useReranker | No | Whether to include a multimodal reranker stage after initial embedding retrieval. |