Evidence Validator

Overview

The Evidence Validator asks a simple but uncomfortable question: what proves this is true?

Lifecycle: Discover through Run
Default autonomy: Analyze and recommend
Human owner: Quality Lead or PMO Lead
Typical consulting roles: Quality manager, PMO lead, audit and controls specialist, workstream lead; typically manager to senior manager

The implementation challenge

Transformation reporting contains many statements that sound precise but are based on outdated documents, incomplete tests or verbal assurance. That is a weak basis for a design gate, cutover decision or audit.

Mission and boundaries

The agent traces material claims to approved decisions, system records, test results and accountable owners. It assesses freshness, authority and completeness and highlights unsupported conclusions.

It does not decide whether weak evidence is acceptable or approve a gate.

Role across SAP Activate

• Discover: validate assumptions behind scope, value and readiness.
• Prepare: check that mobilization decisions have owners and source evidence.
• Explore: verify that fit-to-standard outcomes are captured and approved.
• Realize: validate completion, testing and control claims.
• Deploy: assemble and challenge cutover and go-live evidence.
• Run: support operational assurance and benefit validation.

Key use cases

• trace a status statement to its source.
• identify expired approvals and obsolete test evidence.
• distinguish observed facts from estimates and opinions.
• test whether a gate pack supports its recommendation.
• prepare an evidence trail for audit or independent assurance.

Required skills

Evidence classification, SAP implementation governance, document and record analysis, testing, auditability, provenance, quality management and domain-specific acceptance criteria.

How it works

The agent receives claims, artifacts, approvals, system records and quality criteria. It returns trace links, evidence-strength assessments, missing proof and conflicts. Every conclusion should retain its source and time context.

Human–agent operating model

The agent performs the laborious tracing and comparison. Workstream leads explain context. Quality leaders define what constitutes sufficient evidence. Named decision authorities accept, reject or conditionally accept the conclusion.

Governance and risks

Access to evidence must follow project and data-classification rules. The agent must not turn absence of accessible evidence into an assertion that no evidence exists. Confidence, source quality and limitations should remain visible.

Success and maturity

Useful measures include traceability coverage, unsupported claims found before gates, stale evidence detected, review time and audit findings.

Example

A workstream reports that integration testing is complete. The agent finds that two critical interfaces passed only an earlier build and that one retest remains open. It corrects the evidence picture; the Test Manager determines the response.

Related agents

Transformation Orchestrator, Testing & Quality Agent, Cutover & Readiness Agent, Cross-Artifact Consistency Checker and Quality-Gate Assessor.