Skip to main content

​
Pipelines

Pipelines let you chain multiple analysis steps into automated workflows. You can build pipelines visually in the pipeline builder, or let the AI agent generate one from a natural language description.

​
What is a pipeline?

A pipeline is an ordered sequence of processing steps that run on your data. Each step can be:
  • Client-side — image processing operations that run in your browser (thresholding, filtering, transforms)
  • Server-side — model execution on GPU (SAM3 segmentation, DeepLabCut tracking, etc.)
Steps are connected so that the output of one step feeds into the next.

​
Example pipelines

​
Cell counting pipeline

  1. Threshold (client) → binarize the image
  2. SAM3 segmentation (server) → detect all cells
  3. Count & measure (client) → count cells and compute statistics

​
Animal tracking pipeline

  1. DeepLabCut (server) → detect keypoints per frame
  2. Export CSV (client) → tabular coordinates for downstream analysis

​
Calcium imaging pipeline

  1. Suite2p (server) → motion correction, ROI detection, spike deconvolution
  2. Visualization (client) → plot fluorescence traces

​
Pipeline pane

Pipelines run in a dedicated Pipeline Builder pane type. You can:
  • Switch any pane to pipeline builder mode
  • Drag and drop steps to reorder them
  • Configure parameters for each step
  • Run the pipeline and view results inline
See Building Pipelines for a step-by-step guide.