Skip to content
AI Agents

Map new sources in hours instead of lengthy manual mapping cycles.

Onboarding a new provider, fund or system usually means someone manually working out which column means what. Datox agents map source data from the source structure and your existing reporting model, then the governed transformation rules run once a person confirms the mapping.

How it works

From an unfamiliar file to a working mapping

  1. Analyse the source

    The agent inspects structure, headers, data types, value patterns and sample records to understand what each column represents.

  2. Propose the mapping

    Fields are matched to the reporting model, with alternatives and a rationale where the match is uncertain.

  3. Confirm and adjust

    A member of the reporting team reviews the proposal, corrects anything wrong and approves it. The confirmed mapping becomes configuration.

  4. Apply and reuse

    Transformations, normalisation and calculations run against the mapping every period, and it can be reused for similar sources.

Division of work

Who does what in a mapping

AI assists with interpretation and setup. Configured logic performs the arithmetic. People confirm and approve.

AI agents can

Analyse unfamiliar source structures, suggest field mappings, identify likely relationships and assist with transformation setup.

Configured logic does

Perform calculations, apply approved transformation rules and produce repeatable results every period.

People decide

Confirm mappings where required, review exceptions and approve material changes before they take effect.

Explainability

A transformation you can read back

Agents apply the logic you configured; they do not invent arithmetic. When a reviewer asks what happened to a value, the answer is a description of the concrete steps applied, with inputs.

Typical explanation content

  • Which source field or fields fed the reported value.
  • Which normalisation was applied, currency, date, identifier or classification.
  • Which calculation ran and with what inputs and rate.
  • Which version of the mapping was in effect for that period.
Reuse

Mapping work compounds

Because mappings are configuration rather than code embedded in a script, they accumulate as a reusable asset. The tenth fund from the same administrator is far quicker to onboard than the first.

  • Provider-level mapping templates reused across funds and entities.
  • Shared reporting model across regulatory and internal outputs.
  • Changes applied once and inherited where the mapping is shared.
  • Versioning keeps prior periods reproducible after a change.