> ## Documentation Index
> Fetch the complete documentation index at: https://docs.btsscorp.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Inside an entity

> Source data, mappings, preview, run, and ORCA Learnings

Open an entity to reach its **workspace** — the room where one entity's work happens, laid out as numbered sections you work top to bottom.

<Steps>
  <Step title="Source Data">
    Upload the source extract (CSV, Excel, HDL, or a zip). If the template ships an MVT, use **Download Template**, fill it in, and ORCA auto-maps the fields.
  </Step>

  <Step title="Field Mappings">
    Map source columns to Oracle's import components. MVT fields auto-map; **Propose Mappings** drafts the rest; **Import from cycle** pulls a previous cycle's mappings forward.
  </Step>

  <Step title="HDL Preview">
    Inspect the exact Oracle-format `.dat` ORCA will generate, before anything is sent. Download it to eyeball it.
  </Step>

  <Step title="Run">
    Submit to Oracle — **Run** / **Load to Oracle** for HDL, or the UCM → Load → Import → Purge pipeline for FBDI.
  </Step>

  <Step title="Load History">
    Every load's results and errors, kept per entity.
  </Step>

  <Step title="ORCA Learnings">
    Confirmed error fixes, in three tiers: **Silent** (auto-handled), **Auto-fix** (applied automatically), and **Ask** (needs confirmation).
  </Step>
</Steps>

## Never answer the same question twice

**ORCA Learnings** is what makes each cycle faster. Once a fix is confirmed, ORCA captures it with provenance and reapplies it automatically in later cycles — so it never asks you the same question twice. The section starts empty: *"No learned error patterns yet. ORCA learns from confirmed fixes during chat."*

<Tip>
  Field **mappings** carry forward when you click **Import from cycle** — opt-in per entity, with a diff preview. The **learnings** carry forward automatically.
</Tip>