Banking analytics modernization
Step 1: Ingestion & normalization
Agent configuration
Snapshot of the agent profile for this stage.
Agent: Extractor Agent
Goal: Transform all raw customer files into clean structured and searchable data.
Input: Spreadsheets, PDFs, text exports
Expected output: Unified staging tables and normalized document chunks
System prompt guidance: Prioritize data lineage and flag malformed rows. Preserve source references for every extracted entity.
Current status
Live overview of this step execution.
Step state:completed
Tools registered: 3
Activity events: 3
This is a base observability view. In future iterations it can stream real-time logs and tool outputs.
Tool calls
Each call made (or queued) by the step agent.
| Tool | Purpose | Timestamp (UTC) | Status |
|---|---|---|---|
| spreadsheet-parser | Read XLSX billing inventories and normalize columns. | 2026-03-03T09:11:00Z | completed |
| pdf-ocr | Extract architecture notes and capacity constraints from PDF diagrams. | 2026-03-03T09:14:12Z | completed |
| sqlite-loader | Load normalized datasets into staging tables. | 2026-03-03T09:16:20Z | completed |
Agent activity log
Human-readable log trail for this step.
Detected 6 source files: 2 XLSX, 1 CSV, 2 PDF, 1 TXT.
09:10:54
Mapped 98% of spreadsheet headers to canonical schema.
09:15:31
Flagged 3 rows with missing region data for manual review.
09:17:05