Strand AI Connector for Claude
Spatial proteomics from H&E, right inside Claude
Lattice predicts spatially resolved protein-marker maps from routine H&E, delivered as virtual mIF channels. Add the public Strand AI connector from Claude's Connectors Directory to inspect whole-slide images, run predictions, and interrogate single-cell spatial biology in the conversation.
View Results in Claude
When you open a sample in claude.ai or Claude Desktop, its newest Lattice result appears in the conversation in a read-only viewer. Pan across whole-slide biopsy geometry, zoom to cellular resolution, toggle predicted protein channels in immunofluorescence or brightfield IHC-style rendering, and overlay single-cell nuclei segmentation—or jump into the full Strand AI workspace with one click.
Result images stay in the viewer and are never sent to the model. Under the hood, Claude uses list_samples and get_sample to locate your organization-scoped slide, then opens it in the embedded viewer with strand_ai.view_sample (The sample ID is required). If a presigned transfer expires during a chat upload, Claude can create a fresh upload_sample session and send the bytes directly to storage.
From routine H&E to single-cell spatial reasoning.
Because Lattice outputs single-cell spatial coordinates, morphology measurements, and per-cell protein marker expression alongside full-resolution OME-Zarr pyramids, Claude and other agents can move straight from visual inspection to quantitative spatial biology across hundreds of archived trial biopsies.
Spatial neighborhood analysis
Quantify immune infiltration, tumour-stroma boundary gradients, and cell-to-cell proximity networks directly from predicted CD3e, CD8, CD20, CD68, FoxP3, and PanCK maps.
Cell clustering & phenotyping
Join per-cell morphology Parquet tables with predicted marker expression by instance_id to cluster cell states and compare microenvironments across treatment arms.
Cohort-scale biomarker discovery
Query and stratify retrospective H&E archives in plain language without cutting a new section or consuming scarce biopsy tissue.
Built for translational teams and agent workflows.
The same remote MCP endpoint powers interactive exploration in Claude web and desktop and automated analysis pipelines in Claude Code.
- 01
Connect your organization
Add Strand AI from Claude's Connectors Directory, sign in with OAuth, and choose the Strand AI organization your session acts in.
- 02
Predict & segment
Upload de-identified H&E slides in the browser or from chat, run credit-free nuclei segmentation, and review the credit estimate before starting a Lattice run.
- 03
Explore in chat or export
Inspect multi-channel virtual mIF overlays right in the conversation, or export native OME-Zarr, OME-TIFF, and single-cell Parquet tables for downstream pipelines.
Enterprise-grade data custody
Every connector request is authenticated against your live Strand AI organization and member permissions. Whole-slide uploads stream directly to Google Cloud Storage, viewer tiles stream directly from Strand AI to the embedded viewer, and paid runs require an explicit credit estimate before reservation.
Slides must be de-identified before upload. Automated de-identification is off by default and runs only when enabled for your organization. Do not upload protected health information or other regulated identifiable health data. Step-by-step setup, Claude Enterprise network egress configuration, API key usage, and troubleshooting live in the Strand AI connector documentation. See the Privacy Policy, Terms of Service, and Security and data handling, or contact support@strandai.com.
Research use only (RUO). Not for use in clinical or diagnostic procedures.