Skip to main content

Scientific Knowledge Base

Connect images, extracted features, tables, methods, and documents so a result can be understood in its research context. A canopy-cover measurement is more useful when you can find its source image, processing method, and plot treatment.

Build the evidence collection​

  • Organize datasets and plot assets in the Dataset Editor.
  • Extract measurements with versioned LgoPy blocks.
  • Retain figures, tables, and other results with their runs.
  • Keep PDFs, protocols, manuscripts, and notes with their project or study.
  • Review pipeline definitions, parameters, and logs when tracing a method.

This collection supports comparisons across datasets, further analysis, and research writing while preserving the link between measurements and their origin.

A foundation for multimodal RAG​

Multimodal retrieval-augmented generation (RAG) brings relevant visual, numerical, and written evidence into an assistant's context before it answers a question. PhenoWorks's connected data and outputs provide a foundation for that workflow.

Current retrieval uses the available MCP tools, including dataset summaries, block inspection, and artifact retrieval. Uploading a document does not automatically make its contents searchable by the agent. Retrieval across all images and documents requires suitable indexing and connected tools.

Put the evidence to work​

Ask the Agent to interpret a specific run or draft a section from selected measurements. Identify the dataset and methods, and supply any context it cannot retrieve. For comparisons, check that units, processing settings, and acquisition conditions support the interpretation you want to make.