Platform quickstart

Use the web platform for the standard Strand AI workflow. No API key or local software is required.

​

1. Sign in and choose an organization

Open app.strandai.com. Follow your Strand invitation, or use Continue with Google or Continue with email on the Get started free page. Your active organization owns the slides, credits, and runs you create. If you belong to more than one organization, open the organization menu and choose Switch org before uploading.

​

2. Upload an H&E slide

Select Upload in the platform sidebar to open Upload Sample. Drag an TIFF, TIF, SVS, NDPI, SCN, or MRXS file onto Drag and drop your slides here, or click the drop zone to browse files. Upload only data that your organization is authorized to process. Slides must be de-identified before upload. Do not upload protected health information. See Security and data handling. Run cell segmentation is selected by default. Segmentation finds individual cells in the H&E and costs no credits. It is not billed like a marker run. Keep it selected for per-cell readouts. Clear it if you only need marker channels. Segmentation can be started or retried later from an SDK or MCP client. Segmentation produces viewer overlays and three artifacts on the Export tab: a label mask, a per-cell morphology table, and a manifest. Per-marker cell expression (mean and median per cell) is available through the REST API, the SDKs, and MCP download_results. The AnnData and SpatialExperiment objects built from a prediction are pixel-grid marker matrices, not per-cell tables. Multiple files queue automatically, with up to two uploads active at once.

​

3. Wait for ingest and confirm slide scale

A single completed upload opens its sample automatically. For a batch, use the Upload complete. Open sample link on each completed row. Ingest preprocessing and the H&E pyramid begin automatically after the final byte is committed. Automated de-identification is off by default and runs during ingest only when enabled for your organization. The sample page shows Slide scale confirmation required only when the embedded physical scale is absent or conflicts with the objective metadata. Choose Use slide tag, accept the suggested Use … µm/px value, or choose Enter manually after checking the scanner metadata. A valid scale appears inline and is ready to use. Scale is a property of the sample, not of a run. You can correct it at any point before submitting, and a corrected value applies to every later run on that sample. See Set slide pixel size for the API, SDK, and command-line equivalents.

​

4. Choose markers and review cost

From the sample page, choose Submit a run or New run to open Run Lattice.

1. Review the sample and Slide Scale cards.
2. Choose channels under Select Markers. The picker offers only markers available to the active organization and identifies completed coverage.
3. Review Cost estimate. It shows Patches × Markers, Estimated credits, Org balance, and Balance after submit.
4. Select Run Lattice (… markers) to charge the displayed credits and submit the run.

The button remains unavailable when no marker is selected, scale confirmation is required, or the organization does not have enough credits.

​

5. Monitor the run

Submission returns to the sample’s Markers tab. The page reports queued and running state, completed marker count, partial delivery, or failure. You can leave the page. The run continues server-side, and the platform posts a notification when it finishes.

​

6. Inspect and download results

Open Viewer to compare the original H&E with delivered marker layers. Use the viewer controls to toggle markers, change colors and intensity, switch views, and inspect cell-segmentation overlays when available. See Viewer keyboard shortcuts for navigation controls. Open Export for downloads:

• use the marker download controls for individual OME-Zarr marker pyramids;
• choose Generate OME-TIFF, then Download, for a run-scoped pyramidal OME-TIFF after a prediction completes; and
• download segmentation layers when the sample includes them.

OME-Zarr is the stored format, so those downloads start immediately. An OME-TIFF is rendered on request: Generate OME-TIFF starts the render and Download stays unavailable until it finishes, which takes minutes on a large slide. The rendered file is cached, so a second request for the same run returns the existing file. The archive contains the run’s predicted marker channels as a pyramidal image and includes the source H&E. Segmentation layers download separately.

​

Developer workflows