Back
LIVE DEMO · ILLUSTRATIVE

Backbone module · every gate

Thirty days. Two systems. Zero faith required.

Before any cutover, the Backbone runs in parallel with your existing process — same inputs, both outputs reconciled line-by-line, every day. Discrepancies are root-caused, fixed and backfilled until the register reads zero. You switch on evidence, not hope. Press play, or scrub the thirty days yourself.

Mockup · interactive

Parallel Run — proof before cutover

Day 1 / 30
Day 1 · first runDay 30 · gap report
1 / 30
Parallel run day
173,920
Records reconciled
14
Open discrepancies
97.71%
Line-by-line match rate

Same inputs ⇄ two systems ⇄ reconciled line-by-line, every day

Legacy baseline

SOURCE OF TRUTH
178,000
records processed · ≈180k/day

Your existing process, untouched. It stays in charge until the evidence says switch.

Backbone (parallel)

SHADOW MODE
173,920
matched to baseline · 4,080 lines under investigation

Same inputs, zero production writes — every output line reconciled against the baseline daily.

Discrepancy register

14 OPEN
PR-01Timezone normalisation — legacy stamps in WAT, feed in UTC
~380 lines/day affected · root cause under investigation
OPEN
PR-02Feed mapping gap — card settlement leg missing
~640 lines/day affected · root cause under investigation
OPEN
PR-03Reference-format drift — NIP refs truncated to 16 chars
~300 lines/day affected · root cause under investigation
OPEN
PR-04Suspense scope — legacy excludes GL 9920 sweep lines
~450 lines/day affected · root cause under investigation
OPEN
PR-05Duplicate-suppression rule mismatch on retried NIP credits
~210 lines/day affected · root cause under investigation
OPEN
PR-06Rounding rule — half-up vs banker’s rounding on FX legs
~260 lines/day affected · root cause under investigation
OPEN
PR-07Charge split — VAT broken out on a separate line
~140 lines/day affected · root cause under investigation
OPEN
PR-08Reversal pairing — same-day reversals netted, baseline shows gross
~310 lines/day affected · root cause under investigation
OPEN
PR-09Cut-off window difference — EOD at 22:00 vs 23:59
~520 lines/day affected · root cause under investigation
OPEN
PR-10Weekend batch sequencing — Monday double-batch out of order
~220 lines/day affected · root cause under investigation
OPEN
PR-11Stale BVN cache on the onboarding feed
~90 lines/day affected · root cause under investigation
OPEN
PR-12Encoding — customer names with diacritics transliterated
~120 lines/day affected · root cause under investigation
OPEN
PR-13Multi-leg card-scheme settlements grouped differently
~180 lines/day affected · root cause under investigation
OPEN
PR-14Interest accrual day-count — ACT/360 vs ACT/365
~260 lines/day affected · root cause under investigation
OPEN

Cutover confidence

Gate opens at 95% — keep the run going

Computed from discrepancies resolved and volume reconciled. Nobody argues with this number in a steering meeting.

The rule of the parallel run

  • Same inputs feed both systems for the full thirty days — the Backbone writes nothing to production.
  • Every output line is reconciled against the baseline daily; every difference gets a root cause, a fix and a backfill.
  • The gate opens only when the gap report reads zero unexplained differences across the full run.

Scrub the timeline backwards and items re-open — the register replays from the day, exactly like the evidence pack.

← Back to the demo suite

Illustrative mockup for demonstration. Sample data — not a live production system.