The workflow before
Kane's accounting team reconciled transactions between two platforms, Physical Ops and CMR Ops, by cross-referencing millions of records in Excel with VLOOKUP-style matching. Currency differences were converted by hand, which produced mismatches and false positives that each had to be chased down. Aged, unmatched transactions surfaced late.
What we built
We built a cloud reconciliation engine. Apache NiFi extracts records from both platforms, applies currency conversion automatically, and loads them into Snowflake, where matching rules run on the full data set instead of spreadsheet tabs.
Metabase dashboards show the team what is left: reconciliation mismatches and aged transactions, so accountants work only the exceptions.
We also laid out a phase 2 plan to expose the clean, reconciled data to Kane's own clients as a paid service.
Before and after
| Measure | Before | After |
|---|---|---|
| Reconciliation cycle | About 3 days of manual Excel work | Under 4 hours, mostly exception review |
| Currency conversion | By hand, frequent mismatches | Automatic in the pipeline |
| Aged transaction visibility | Found at period end | Daily dashboard |
What's next
The matched data in Snowflake is the base for an L3 workflow agent that triages the remaining exceptions, proposes the likely match and routes only true breaks to an accountant.
