AI Agents
Discovery Express

Finance & back office

Master Data Agent

Finds the duplicate customer, the wrong tax id and the missing field before they reach a report.

Reference architectureReviewed September 3, 2026
Internal apps · Master Data
Demo data

  1. ERP

    New supplier created: DISTRIBUIDORA DEL BAJIO SA DE CV.

  2. Agent

    Same tax id as a supplier created in 2019 under a different spelling, and both have open purchase orders. I am not proposing a merge while transactions are open.

  3. Duplicate flagged with confidence 0.96 · evidence attached · no merge, no deletion, nothing overwritten

  4. Handoff to a person

    The data owner closes the open orders and approves the merge. Every merge stays reversible with its before value stored.

Internal appsSAPMaster data hubSalesforce
KPI it moves
Duplicates down 80 to 95%Typical target
Channels
Internal appsSlack / Teams
Integrates
  • SAP
  • Master data hub
  • Salesforce
  • Snowflake
  • Your ERP
A data steward comparing two printed customer records with a highlighter
Finance & back office

What it does

The duplicate, the wrong tax ID and the missing field, caught before a report.

Where it lives

Your cloud, your repo, your IP.

1. Receives 2. Looks up 3. Decides 4. Acts 5. Escalates

Systems it integrates

  • SAP
  • Master data hub
  • Salesforce
  • Snowflake
  • Your ERP

Master Data Agent

03Decides

Channels

  • Internal apps
  • Slack / Teams

A person on your team

01

Receives

Watches new and changed records in every system that creates a customer, supplier or product.

02

Looks up

Compares against the golden record, the public registry data and your naming standard.

03

Decides

Judges whether two records are the same entity and says with what confidence and why.

04

Acts

Completes the missing fields it can prove and proposes the merge for the data owner to approve.

05

Escalates

Sends every low confidence match and every entity with open transactions to a person.

KPI it moves

Typical targets, no result yet.

The first KPI measured in a client pilot is what promotes the label.

Proof Ledger
Typical targets

Typical target

80 to 95% fewer

Duplicate records

Typical target for a reference architecture, not a result. DaCodes, Sep 2026

TargetNo result yet

Typical target

95% or more

Records complete on the required fields

Typical target for a reference architecture, not a result. DaCodes, Sep 2026

TargetNo result yet

Typical target

Half or less

Hours of manual cleanup per month

Typical target for a reference architecture, not a result. DaCodes, Sep 2026

No result yet

Before

Baseline

Target

Half or less

Reviewed September 3, 2026

How it is governed

Every action has an owner, an eval and a limit.

Set in the Blueprint, enforced by the Govern layer of the platform.

Govern · Approval queue
Demo data

Waiting for a person

Merge 2 supplier records with the same tax ID · the data owner decides

Open orders closed first

Weekly eval

Passing

False merge rate

0%

Precision against a manually matched set within target

Limit, enforced

Every change keeps its before value

Human approvals

  • Every merge, by the data owner
  • Changes to tax or banking fields, by finance

Typical evals

  • Precision and recall against a manually matched set
  • False merge rate, which must stay at zero

Limits

  • It never merges on its own and never deletes
  • No changes to records with open transactions

Time to production

8 to 12 weeks

of pilot to production under a KPI Contract, after a Blueprint of 1 to 2 weeks

As a reference architecture, plan for the upper end of the pilot range: integration to your systems takes most of the time, and the first measured KPI is what promotes the label.

DaCodes delivery standard, Sep 2026. Integration is the long pole, not the model.

How we deploy
The four phases
Commitment gate

  1. 01Discovery Express

    2 sessions, no cost

  2. 02Blueprint

    1 to 2 weeks

  3. 03Pilot

    8 to 12 weeks

  4. 04Scale & Managed

    Continuous

  • Pilot, first 8 weeks
  • Up to week 12 of the pilot
  • This agent: plan for the upper end
  • Commitment gate: the KPI decides Scale

Proof and demo

What backs the label.

Proof Ledger note

Reference architecture

No result yet

No client pilot has closed with a measured KPI. The first one promotes the label.

Reviewed September 3, 2026

Where this agent sits on the maturity policy

  1. Reference architectureToday
  2. Pilot
  3. In production

The Discovery Express shows this agent on your data, live. Two sessions, no cost.

See this agent with your data

Related agents

Same function or industry.

All agents
A fraud analyst at two monitors at night above Paseo de la Reforma
In production

Finance & back office

Fraud Review Agent

KPI it movesMeasured

Review time down 55% to 70%

FAQ

Direct answers.

No. It proposes, with the evidence and the confidence. The data owner approves, and every merge stays reversible with its before value stored.

No. It can work directly against your ERP and CRM. If you do have a hub, it writes there and leaves the source systems alone.

No client pilot has closed with a measured KPI. The figures here are typical targets, and the first measured pilot promotes the label.

Which process do we automate first?

Two discovery sessions, no cost, and you leave with two or three quick wins whether or not you build with us.

We reply in under one business day.

Loading the form