Overview
Strand AI builds AI models that predict missing biological readouts from the data researchers already have. The first capability available through this connector is Lattice, which predicts spatially resolved protein-marker channels (virtual multiplex immunofluorescence) from routine H&E histology slides, so teams can recover spatial proteomics from archival tissue without extra staining or sample consumption. In Claude, researchers can upload a de-identified whole-slide image, browse samples and available markers, estimate credits before submitting, run or cancel an inference job, request cell segmentation, and download OME-Zarr or OME-TIFF results. OAuth binds each connection to one Strand AI user and organization. Reads, marker discovery, segmentation, and dry-run estimates are credit-free; a real submission reserves the displayed estimate. Slide and result bytes transfer directly between the client and Strand AI-managed storage. Slides must be de-identified. Research use only. Not for clinical or diagnostic procedures.
