Martenweave · local migration evidence

Know what needs a decision before the next mock load.

Drop in one migration file. Martenweave keeps it local, tests it against governed model context, and turns scattered exceptions into recorded decisions before change approval.

Run the evidence case See the four steps

Apache 2.0 local-first deterministic checks human approval

Martenweave Work Canvas showing 13 reviewable findings, four decision groups, and a human disposition panel for a local customer migration file
Verified with the checked-in synthetic customer migration sample. No customer data is shown.

Your source stays local.The run records its file hash and evidence boundary.

Rules expose the gaps.Findings remain separate from assumptions and review state.

Humans decide first.Approval stays blocked while evidence is unreviewed or deferred.

One command · one evidence case

Start from the artefact the team already has.

Use CSV, XLSX, XML, or JSON. The SAP BP Customer starter gives the run enough governed context to separate structural evidence from decisions that still need a person.

$ python -m pip install martenweave-core
$ martenweave start ./customer_messy.csv \
    --template sap_bp_customer_migration

verdict: blocked
reviewable findings: 13
decision groups: 4
classified: 0 / 13
canonical changes: 0
next: classify one finding

Run it locally →Inspect the source →

The product

An evidence workspace, not another migration dashboard.

Core preserves canonical truth. The Local API exposes governed capabilities. Workbench keeps evidence, context, and the next human decision in one minimal canvas.

01

Evidence case

One source artefact, its hash, deterministic findings, assumptions, and review state stay connected.

02

Human disposition

Confirm, exclude, accept risk, or defer each finding. Accepted risk and deferral require a durable rationale.

03

Evidence-gated change

The candidate proposal cannot be accepted while findings are unreviewed or deferred; apply remains a separate explicit action.

The loop

File → evidence → disposition → gated change.

The fastest path to value stays understandable enough to defend in a migration review.

  1. 01 · StartCreate a local evidence workspace from one representative file.
  2. 02 · InspectGroup deterministic findings into the decisions that block readiness.
  3. 03 · DecideRecord a human disposition and rationale without changing canonical model files.
  4. 04 · ReviewOnly after the evidence gate clears, accept, reject, or return the candidate proposal before apply.

Use cases

Built for the moments when model knowledge becomes operational risk.

SAP migrationConnect source datasets, mappings, target fields, and readiness evidence before testing exposes the mismatch.MDM and MDGPreserve the implementation knowledge around governed operational master data.AMS and changeGive support teams a defensible starting point for investigating and reviewing a model change.

Start narrow

Bring one difficult migration file.

Use a representative Customer or Business Partner slice to prove the evidence and review loop before expanding the model.

Run locally

Consulting

Use Core with an accountable delivery method.

Engagements focus on the model and evidence that need to survive a handover—not on inventing a parallel operating platform.

How consulting works

Where it fits

A supporting layer for the systems you already run.

Martenweave keeps model evidence inspectable around delivery systems. It does not replace their transactions, configuration, workflow, or source of record.

01SAP ECC & S/4HANA

Connect legacy columns, SAP field endpoints, contexts, mappings, and migration evidence before testing.

02SAP MDG & MDM

Keep field meaning, ownership, value rules, and implementation decisions reviewable alongside governed data.

03ERP, CRM & warehouse

Use shared business attributes and lineage to make cross-system definitions and downstream impact explicit.

04Jira, Confluence & delivery files

Turn tickets, documents, workbooks, and incident findings into durable evidence without replacing the tools.

AI workflow

Give AI structured context—then keep the decision under control.

Instead of asking an agent to reconstruct meaning from a chat or a spreadsheet, give it the canonical model, evidence, and boundaries it needs to prepare a focused next step.

  1. 01
    Ground the request

    Use canonical objects, references, source evidence, and known context as the working frame.

  2. 02
    Draft a proposal

    AI can prepare a reviewable PatchProposal from a note, finding, or design question.

  3. 03
    Validate deterministically

    Rules check object shape, references, and supported SAP context before anything is indexed.

  4. 04
    Approve with a human

    Reviewers decide whether a proposal becomes canonical truth and an auditable change.

Product FAQ

Questions teams ask before they put model knowledge under control.

Short answers for migration, MDM, governance, and AI-assisted delivery teams. For detailed product boundaries, see the full FAQ.

What is Martenweave?

Martenweave is an open-source model governance and evidence layer. It turns model knowledge from spreadsheets, datasets, tickets, and SAP context into canonical files that can be validated, traced, reviewed, and exported.

Which systems and projects does Martenweave support?

Martenweave is an open-source core with a tested SAP domain pack and synthetic example models for SAP ECC to S/4HANA migration, MDM/MDG delivery, data governance, and AMS scenarios. It ingests evidence from Jira and Confluence exports, mapping workbooks, and delivery files as inputs to review. It is not certified by SAP or affiliated with SAP, and it does not replace SAP MDG, Migration Cockpit, Syniti, SNP, Jira, or Confluence.

Does Martenweave replace SAP MDG, S/4HANA, Jira, or Confluence?

No. Martenweave is a supporting model-governance layer. It does not replace system transactions, configuration, workflow, source-of-record tools, or direct SAP write-back.

How does Martenweave help SAP migration and MDM teams?

Teams can connect business attributes, source columns, SAP field endpoints, mappings, value rules, ownership, lineage, validation findings, and change proposals in one reviewable model.

How does the AI workflow work?

AI works from canonical objects, references, and source evidence to draft a reviewable PatchProposal. Deterministic validators then check the proposal, and a human decides whether it becomes canonical truth.

Can AI change the model automatically?

No. AI can prepare proposals and explain structured context, but it does not silently mutate canonical model files. Human approval and an auditable change path remain required.

Capacity scenario

Make the cost of hidden rework visible.

Use your own team assumptions to estimate the annual capacity that clearer model evidence and controlled change could recover.

Illustrative annual capacity recovered

€109,4408 people × 3 hours × €95 × 48 working weeksAn editable planning scenario, not a savings guarantee.
Discuss a real baseline