Skip to content
OpeniDevelopers

CasesReconciliation at scale

More than 100 million records reconciled

NDD uses Nexus to reconcile more than 100 million records. Data comes in by file, database or webhook, the rule set matches the two sides and each pair comes back with a decision and a score, with no manual queue for what checks out.

NDD
Request accessDocumentation

From source data to result

  1. 1Data sources
  2. 2Rule set
  3. 3Matching
  4. 4Pairs with a decision
records reconciled
100M+

The challenge. Reconciliation at scale.

When an operation reaches millions of records, reconciling stops being a rules problem and becomes a scale problem. The files are too big for a spreadsheet, the two sides live in different systems and any small exception rate turns into a queue nobody can work through by hand. What solves it is a process that runs on its own for what checks out and hands over only the mismatches.

Two datasets, one rule set
Example
origem_a (file)origem_b (database)

  • Record 000184

    R$ 1,284.90

    MATCH

    Record 77A1

    R$ 1,284.90

  • Record 000185

    R$ 312.00

    MATCH

    Record 77A2

    R$ 312.00

  • Record 000186

    R$ 4,790.15

    POSSIBLE

    Record 77A9

    R$ 4,790.51

  • Record 000187

    R$ 96.40

    NO_MATCH

    no pair

Sample data, to illustrate the flow.

How it works. In 3 steps.

  1. 01

    Sources come in as they are

    Files, database tables and webhook events are registered as data sources. Each one feeds a dataset, with no manual conversion.

  2. 02

    The rule set compares the two sides

    The fields of each dataset become canonical keys, such as amount and document, and the rule defines what counts as a pair and with what score.

  3. 03

    Only the exception goes to a person

    Each pair comes back with a decision and a score. What is MATCH moves on; POSSIBLE and NO_MATCH are set aside for review.

How it runs through the API. In documented calls.

  1. POST /v1/data-sources

    Nexus receives the data source that goes into the reconciliation.

  2. POST /v1/rule-sets/{ruleSetId}/run

    The rule set runs and matches the two datasets.

  3. GET /v1/matches

    Each pair comes back with a decision and a score.

What stays visible. Set apart for the people who operate.

  • Record with no pair on the other side
  • Mismatched amount or date
  • Pair below the minimum score

Other cases

“We integrated Open Finance in one sprint. The documentation answers before we need to ask.”
Marina DuarteHead of Engineering, credit fintech
“The reconciliation that took the team two days now runs on its own.”
Rafael AntunesCTO, retail chain

Teams from fintechs, retail and accounting already integrate with Openi.

Request access

Build your reconciliation. With the same calls.

The cases use documented endpoints only. Request access and run the flow with your own operation's data.

Request accessDocumentation