Nathan Henderson, CPA
Personal site for projects and writing.

Claim timeline reconstructor

Private Sensitive

Turns evidence into a defensible timeline, plus a dashboard for claim status sharing.

What I built

  • Preserved original timestamps with confidence levels.
  • Flagged gaps and delayed reporting without speculation.
  • Separated facts from commentary in the output.
  • Added photo evidence ingestion via filename parsing.
  • Generated dashboards and executive summaries for status sharing.

Context

Insurance support context · 80–120 sources

Stack

Python / PDF/EML/DOCX parsing / CLI tooling / Markdown/HTML outputs

Outcomes

  • Reduced review time by 40–60% while standardizing evidence labeling and provenance.
  • Preserved limited funds for evidence discovery phase when org was navigating a crisis.

Details available on request.

Screenshots and examples use synthetic or redacted data.

Notes

Tooling write-up for organizing evidence. It does not provide legal advice or conclusions.

What this is / Who it’s for

Deterministic reconstruction of insurance claim timelines for internal review, board summaries, and insurer-facing discussions. It preserves what happened, when, and based on which source. No inference, no blame.

TL;DR

Inputs and outputs

Dashboard for sharing claim status

Generates a claims dashboard from action CSVs and internal notes: status, next actions, owners, and last evidence dates. Designed for sharing without mixing in non-evidence commentary.

Photo evidence support

Photo evidence can be ingested from a folder. Filenames with a date (YYYY-MM-DD or YYYYMMDD) and optional claim ID become timeline events tagged as “Photo evidence” with “Inferred” confidence unless verified elsewhere.

CLI usage

claim-timeline run --input <folder> --output <dir>

Common flags: --photos, --actions, --notes, --calls, --keywords, --dates.

Sample output

Claim timeline output sample