ReconzaCheck your mapping

Value-blind reconciliation

Know exactly where healthcare data went.

Validate mappings, reconcile every record, catch pipeline failures before data loss becomes a production incident.

Receives
Counts, hashes, shapes
Never receives
A single patient value

No signup. Column names only — this never asks for patient data.

The reconciliation screen

One number, and every row behind it.

Every discrepancy is grouped by cause and cited to a named rule, so an implementation team fixes the mapping instead of guessing at it.

Run 1182 — ReconciliationSynthetic sample

98.5%

126,599 of 128,441 source keys accounted for.

Missing
1,842
Duplicated
12
Modified
340

Error groups, ranked by rows affected

1,842 patient rows rejected — birthDate matched pattern DD/MM/YYYY, target requires YYYY-MM-DD (Rule PT-004)

12 patient rows duplicated — a weak match key created two records for one MRN (Rule PT-011)

340 rows modified in transformation — coverage_code translated outside the destination value set (Rule PT-018)

Next fix: correct birthDate format to recover 1,842 rows.

Sample reconciliation screen with synthetic demo data: Run 1182 resolved a reconciliation rate of 98.5 percent, with 1,842 rows missing, 12 duplicated and 340 modified, each error group cited to a named rule and ranked by rows affected.

Every Implementation moves through the same five stages

Synthetic sample

  1. 01

    Mapping

    34 columns

  2. 02

    Transformation

    340 modified

  3. 03

    Validation

    3 rules failed

  4. 04

    Load

    126,599 inserted

  5. 05

    Reconciliation

    98.5% accounted for

Keeps running after go-live

Your pipeline posts a manifest after every nightly job to the same endpoint, and Reconza flags schema drift, volume cliffs and rising reject rates.

Signed report

A licensed Implementation produces a dated, versioned Reconciliation Report — the artifact your team hands the hospital at go-live sign-off.

Reconza reconciles without reading.

The Probe runs inside your perimeter and sends a manifest: what a record looks like, never what it says. Counts, key hashes and field shapes, and nothing else.

What the Probe sends

  • Row and column counts
  • Field-level hashes and shapes
  • Rejection and rule codes
  • Timing of each pipeline stage

What Reconza never receives

  • Patient names, MRNs, or DOBs
  • Diagnoses, notes, or lab values
  • Any clinical or financial value
  • A copy of the record itself

One license per implementation. No subscription.

The free reconciliation shows the top-line number and the error groups on your own manifest. The license adds the drill-down, the Copilot, and the signed report.

Implementation License

$1,900

One-time, per implementation

  • Unlimited further runs
  • Drill-down to affected key hashes
  • Copilot answers, grounded in your diff rows
  • Signed, dated Reconciliation Report
  • Alerts and post-go-live drift monitoring
Check your mapping first

Implementation License 5-pack

$7,900

For teams running a pipeline of go-lives

  • Everything in the Implementation License
  • 5 implementations, used as you onboard them
  • Same value-blind Probe, same signed report
  • No seats, no usage meter
Check your mapping first

Checkout is hosted by Stripe; NanoCorp is merchant of record.

Read the Probe before you run it.

The Probe is one self-contained script that runs inside your own environment. Before it sends anything, it prints exactly what it will send — counts, key hashes, field profiles, never a value — and waits for you to confirm.

That is why a hospital security review takes an afternoon, not a six-week questionnaire: there is nothing to grant access to.

Probe run — dry preview, nothing sent yet
patients.csv        128,441 rows   sha256 keys: 128,441
encounters.csv       94,208 rows   sha256 keys: 94,208
birthDate            format DD/MM/YYYY   12,940 rows
Confirm and send manifest? [y/N]

Illustrative, abridged — the real Probe output is longer.