Skip to main content
Request Demo

SAP AI Agent · Master data · PP · MM Demo available

Monitor and fix SAP master data health with governed AI agents.

Check your SAP master data every morning. The agent scores the health of materials, BOMs, routings and other objects, estimates what each problem costs, and hands each batch of problems to its owner with an evidence-backed fix to approve. It starts in shadow mode and never changes SAP without the owner's decision.
Request demoSee a worked example
  • Human-approved
  • ECC & S/4HANA
  • SOC 2 Type II
  • ISO 27001
Material master health · morning checkagent active
  1. 01Daily check runMaterials, plant data, BOMs, routingsDone
  2. 02Health scored66.4 out of 100 across 5,013 materialsAt risk
  3. 03Problems batched by ownerMRP controller missing at plant 1010: 20 materialsBatched
  4. 04Fix proposed with evidenceFrom similar materials · 0.99 agreementProposed
  5. 05Owner decisionPurchasing and MRP reviewingWaiting

Illustrative example · not a production transaction

Triggered by
Daily schedule or on demand
Reads
Materials, plant data, BOMs, routings
Validates against
Your rules and similar records
Writes to SAP
Agreed fixes only, once enabled
Approved by
Data owner per area
Runs on
SAP ECC · SAP S/4HANA

Why it matters

Bad master data is the root cause of problems everywhere else.

Missing MRP settings, wrong valuation data and records nobody uses cause failed MRP runs, wrong costs and blocked transactions, and one-off cleanup projects never last.

  • Hidden gaps

    Missing settings nobody sees

    A missing MRP controller or valuation class doesn't fail loudly. It quietly breaks planning and costing.No planner → Missed exceptions → Shortages
  • No owner

    Everyone's problem, nobody's job

    Data issues span planning, purchasing, finance and sales, so they wait for a cleanup project.Workarounds → Spreadsheets → Bad decisions
  • One-off cleanups

    Clean once, dirty again

    Projects fix the data, then the same problems return because nothing watches it day to day.Rework → Migration risk → Lost trust

How it works

Read. Validate. Decide. Act.

The same four steps as every Artificio agent. The agent does the research and the paperwork; your teams make the decisions.

  1. 01 · Read

    Run the checks

    Every morning or on demand, from SAP data feeds.

    • Materials, plant data, BOMs, routings and work centers
    • Cross-object checks, such as subcontracting with no BOM
  2. 02 · Validate

    Score and value

    Scores each material and plant and estimates what each problem costs.

    • Problems weighted by severity and open demand
    • Each amount shows how it was worked out
  3. 03 · Decide

    Group and propose

    Batches problems by owner and proposes fixes from similar records.

    • Proposals from similar materials, with agreement scores
    • Never invents a value; uncertain rows left to the owner
  4. 04 · Act

    Owner decides

    Owners agree, change or reject each fix.

    • Agreed fixes written to SAP once enabled
    • Fixes confirmed by the next data download

Worked example

One batch of problems, from morning check to owner decision.

The morning check finds 20 materials at plant 1010 with no MRP controller. Here is what the agent puts in front of the owner.

Batch · MRP controller missing at 101020 materials
Owner
Purchasing and MRP
Plant
1010
Materials
20 active ROH materials
Common settings
MRP type PD, procurement type F
Open demand
Open requirements on all 20
Worth fixing
USD 3,000 estimated
Planning exceptions reach no planner20 materials

Without an MRP controller, exception messages for these materials are not assigned to anyone.

Proposed fixes · from similar materialsProposed
Product groupProposedWhy
Group 001MRP controller 001Similar materials in group 001 at plant 1010 use 001 · 0.99 agreement
Group 002MRP controller 0010.99 agreement with similar materials
Group 003MRP controller 0020.99 agreement with similar materials
Group 005MRP controller 0030.99 agreement with similar materials
ExcludedNoneNo row is marked for deletion or blocked
  • ProposalsAgentReady
  • Owner decisionPurchasing and MRPWaiting
  • Write to SAPAgentShadow mode

In shadow mode, agreed fixes are recorded but not written to SAP. The agreement rate shows when the agent is ready for more.

Illustrative example · names and values are fictional and depend on your SAP configuration

Inside the platform

Every case is traceable from start to finish.

The case shows who it is waiting for, what happens next, what the agent checked and exactly what each person is approving.

MRP controller is missing at 1010: 20 materials
Purchasing and MRP · started Thu 29 Oct, 22:19 · agent version v1
Waiting for ownerDocumentsAuditRe-assess
CheckedBatchedWith its ownerConfirmed
Now

Waiting for A. Chen (Purchasing and MRP) to answer, asked Thu 29 Oct at 22:21.

Next

A reminder goes out tomorrow at 21:21 if there is no answer.

MaterialPlantProposal
100131Plant 1010MRP ctrl 001
100138Plant 1010MRP ctrl 001
100152Plant 1010MRP ctrl 002
100167Plant 1010MRP ctrl 003
Question for A. Chen

Apply the proposed MRP controllers to all 20 materials at plant 1010? Each proposal comes from similar materials in the same product group.

Agree allChange rowsAnswer as authorized delegate
  • If a fix needs a person in SAP

    Said plainly

    Retiring a material, merging duplicates or anything involving a deletion flag is agreed with the owner and done in SAP by a person.
  • If a data feed is out of date

    Not checked, not guessed

    Checks whose data isn't current are marked as not run, so problems are neither found nor closed on old data.
  • If a fix doesn't hold

    Reopened after the download

    Fixes are confirmed by the next data download. If the problem is still there, the case says who to ask next.

Recreation of the Artificio case screen · names and values are fictional

Capabilities

Master data quality with governance built in.

Continuous checks and owner-approved fixes, instead of one-off cleanup projects.

  • Score

    Health score

    Each material and plant scored, rolled up by how much each is used.
  • Value

    Cost of each problem

    Estimated exposure per problem type, with how the amount was worked out.
  • Check

    Cross-object checks

    Materials, BOMs, routings, work centers and production versions checked together.
  • Route

    Owner batches

    Problems grouped by owner and plant, with reminders.
  • Propose

    Evidence-backed fixes

    Proposals from similar records, with agreement scores and exclusions.
  • Learn

    Rules you approve

    Rules found in your data are approved by the data steward before use.

Ownership

Each problem goes to the team that owns the data.

You name an owner for each data area. Whoever answers first decides, and anything without an owner goes to the data steward.

  • Basic data and units

    Basic data

    Material type, group, units, weights and descriptions.
  • Purchasing and MRP

    Plant data

    Purchasing group, MRP type, lot size and MRP controller.
  • Production scheduling

    Production data

    Backflush, production storage location and scheduler.
  • Accounting

    Valuation

    Valuation class, price control and price.
  • Sales

    Sales data

    Sales organization, distribution channel and item category.
  • Engineering

    BOMs and routings

    Components, operations and work centers.
  • Production planning

    Production versions

    Valid BOM and routing combinations.
  • Data steward

    Exceptions

    Possible duplicates, learned rules, unowned problems and a weekly summary.

Owners, rules and tolerances are configured to your organization and agreed for each deployment.

Controlled autonomy

Start with human oversight. Automate only when ready.

The agent proves itself on your real work before it writes anything to SAP. The platform measures how often your people agree with it, so you decide when to give it more room, case type by case type.

  1. Step 1Where every agent starts

    Shadow mode

    The agent proposes every action but writes nothing to SAP. Your people decide, and each answer counts toward the agreement rate.

  2. Step 2

    Approve, then act

    The required owners approve. The agent acts in SAP only when the last approval lands.

  3. Step 3

    Act by approved rule

    For case types you explicitly configure and approve, such as filling a missing MRP controller where similar materials agree, the agent acts on its own and reports back.

Agreement with the agent · by case typeIllustrative
Missing MRP controller
97%
Missing valuation class
84%
Possible duplicates
71%
  • Approval rules are enforced by the platform, whatever the agent's instructions say
  • Deletion flags and duplicate merges are always done by a person
  • SAP credentials never reach the agent or its instructions
  • Every value carries its source; approved values are locked

Governed agent runtime

Shared platform controls across every Artificio agent.

What makes this agent different is its configuration: its instructions, the tools it may use and the policies it works under. Reviewers, approvals, versions and audit work the same way for every agent you add.

Reviewers
Basic data and unitsmdm@yourco.com
Purchasing and MRPplanning@yourco.com
Accountingfinance@yourco.com
Data stewardsteward@yourco.com

Assigned reviewers

Decisions go to named owners or authorized teams, with reminders, and every email links back to the case.

Tools
Run health checksRead
Propose fixes from similar materialsRead
Record owner decisionRead
Submit agreed fixes to SAPAction

Read tools and Action tools

Gathering data is kept separate from changing SAP. Action tools are controlled and can require human approval before they run.

Instruction versions
v3ActiveHealth check rules updated
v2New plant added
v1Created from blueprint

Versioned instructions

Versions can't be edited, and each case keeps the version it started with, so a change never alters work in progress.

Recreations of Artificio platform screens

What we measure

Results you can see in your own numbers.

Every pilot tracks the same four measures from day one, so the business case comes from your data, not ours.

  • Health score

    Score per plant and overall, tracked over time.

  • Open problems

    Critical, high and medium problems open, by owner.

  • Estimated exposure

    Cost at risk in open problems, with how each amount was worked out.

  • Fixes that hold

    Fixes confirmed by the next data download.

For SAP teams

Integrates with SAP ECC and S/4HANA through standard interfaces.

The agent reads and writes data through supported SAP interfaces, without modifying standard SAP objects and without replacing your existing governance.

Reads from SAP

MaterialsBasic, plant, MRP, purchasing, accounting and sales data
BOMs and routingsComponents, operations and work centers
UsageStock, open demand and receipts, from cached SAP feeds
ExportsSE16N exports (TXT, CSV, XLS) for read-only starts

Writes to SAP

FixesAgreed fixes through supported interfaces, once enabled
Shadow modeAgreed fixes recorded but not written
Change logEvery fix with owner, old value and new value
ResponsesEvery SAP response recorded on the case for resolution or retry

ScopeThe exact interfaces, objects and fields in scope are confirmed for your SAP release and deployment during technical discovery. S/4HANA Cloud Public Edition is reviewed case by case, and some landscapes may need additional configuration.

Pilot approach

Start read-only on one plant.

A focused pilot in your QA or sandbox system proves the agent on your own data before anything reaches production.

  1. Phase 1

    Connect

    Review SAP edition, interfaces, authorizations and representative data.

  2. Phase 2

    Configure

    Agree one plant, the checks in scope and the owners.

  3. Phase 3

    Pilot

    Run daily read-only checks and agree the top ten problems to fix first.

  4. Phase 4

    Expand

    Review the four measures, then add plants, objects and write-back.

Related agents

Prevent bad data, and fix what's already in SAP.

Two connected agents: one stops bad materials getting in, the other cleans up the materials you already have.

This agent

Master Data Health Agent

Problem
Existing master data is incomplete or inconsistent
Value
Problems found, prioritized and fixed by owners
Its partner

Material Master Creation Agent

Problem
Slow, inconsistent creation of new materials
Value
Faster, governed material onboarding
Explore agent

Frequently asked questions

Questions SAP teams ask.

Which master data does it check?
It starts with production planning data: materials, plant data, BOMs, routings, work centers and production versions. Sales, finance and purchasing data can be added.
How is the health score calculated?
Each material at each plant starts at 100 and loses points per problem, weighted by severity and more when it has open demand. Scores roll up by how much each material is used.
Where do the cost estimates come from?
Problems that change planning are valued on the coming weeks of demand at the SAP price; others carry a fixed amount. Each problem shows how its amount was worked out. They are estimates, not booked savings.
Can we start without a live SAP connection?
Yes. Checks can run on SE16N exports in TXT, CSV or Excel format, which is a quick way to start a read-only pilot.
What about duplicates and deletion flags?
The agent identifies possible duplicates and materials marked for deletion but still in use. Retiring or merging materials is agreed with the owner and done in SAP by a person.
Does the Master Data Health Agent write to SAP without approval?
Not by default. Every agent starts in shadow mode, where nothing is written to SAP. After that, the required owners approve before the agent acts. Automatic action is only possible for cases you explicitly configure and approve.
Which SAP versions are supported?
SAP ECC and SAP S/4HANA. The exact interfaces and scope are confirmed for your release and deployment during technical discovery; S/4HANA Cloud Public Edition is reviewed case by case.

Security & compliance

Enterprise security across every agent

ISO 27001:2013 certified, SOC 2 Type 2 compliant, GDPR and HIPAA ready. Every agent action is logged and auditable, and SAP credentials stay in the platform's integration accounts.

  • ISO 27001:2013
  • SOC 2 Type II
  • GDPR ready
  • HIPAA ready

See it working

Two ways to get started.

Exploring

Request demo

See a morning health check, the cost of each problem and an owner batch with proposed fixes, using sample data.

Request demo
Ready to evaluate

Discuss an SAP pilot

Scope a read-only pilot on one plant in your SAP QA or sandbox system, or on SE16N exports, with the four measures agreed up front.

Discuss an SAP pilot