Skip to content

The pipeline

Scelo's core is a three-stage pipeline. It's a guided flow where each stage has a clear job and its own scoped AI assistant.

SOFT Data load · clean · shape intake TOOLS Models pick · suggest compute HARD Results run · board pack

The macro view

The macro view — three pipeline stages wired together

Dashboards → Scelo shows the three stages as cards on a canvas:

  • Each card carries a status summary (rows/cols for Soft, model count for Tools, run count + headline for Hard).
  • A small chat box on each card answers stage-scoped questions without drilling in (e.g. "restore a project", "swap chain-ladder for Mack", "explain this ultimate").
  • open → drills into that stage's full workstation.
  • EXPORT · WHOLE PIPELINE turns the entire flow into a script.
  • + START PROJECT scaffolds a new project.

The three stages

  • Soft Data

    Load, inspect, clean, reshape, and augment a dataset. The intake desk.

  • Tools

    Choose actuarial models — by hand or AI-suggested. The model bench.

  • Hard Data

    Run the models, read the canvas, export the board pack. The readout desk.

State flows forward

The dataset you load in Soft is available to Tools and Hard; the models you pick in Tools drive the runs in Hard. You can move freely between stages and your work persists.