← Back to selected work

03 · Quality & delivery

Wingman & evidence-backed reporting

Detect spreadsheet defects, propose fixes for approval, and reconcile sources before anything is called done.

Evidence map

Recorded residuals, state diffs, gates, and native readback

The workflow compares source truth to output and routes classified differences to accountable owners.

  1. Source baseline
  2. Controlled write
  3. Deterministic validation
  4. Exception route
  5. Human signoff
  6. Native readback
Boundary

Sanitized examples only; client data and non-public implementation details stay private.

Constraint
Complex reporting can look complete while mappings, formulas, links, or presentation remain wrong—and manual review does not scale.
Built
Wingman: Chrome extension + Python service that scans live Workiva sheets via API, flags defects (formula hygiene, broken refs, linked blanks), explains each issue, and applies narrow safe fixes only after confirm-before-write. Alongside: source-first reconciliation, controlled write batches, deterministic validators, and human signoff on customer-impacting changes—used across financial reporting and public-sector implementations.
Role
Designed and built Wingman; designed and operated the delivery workflow across enterprise SaaS implementations for 12+ years.
Human boundary
Automation detects, classifies, and routes exceptions; accountable owners resolve classified differences and approve consequential changes.
Evidence
Logged fix outcomes improving detection over time; formula/formatting gates; prestate/poststate diffs; native source verification.
Scope
Examples are sanitized. Client data and non-public implementation details are not exposed.
Get in touch