Skip to content
Jiandong Pei

SAP BTP & Enterprise AI Architecture — Boundary Model

Work With Me

Independent architecture judgment for SAP BTP and enterprise AI decisions where local implementation success no longer proves that the structure is safe to evolve.

This page is for fixed-scope external architecture judgment. If your organization is evaluating whether this capability belongs inside an internal architecture, platform, or AI governance team, see For Architecture Teams.

When to engage

The right time is when the team can still make the system work, but no one can prove whether the architecture deserves to proceed as designed.

SAP BTP extension decisions

  • Postman works, Work Zone / AppRouter fails.
  • One tenant works, the next tenant breaks.
  • S/4 data has become shadow truth.
  • Clean Core direction is approved, but extension responsibility remains unclear.
  • The system runs, but lifecycle change requires manual repair.

Enterprise AI decisions

  • An agent can execute, but no one can prove whether it is safe to act.
  • Agent, workflow, extension, and core-system responsibility is blurred.
  • RAG, tools, workflow state, identity, and authorization are crossing boundaries.
  • Prompt, tool, policy, model, or context changes cannot be reconstructed.
Strategic decision

Strategic Boundary Verdict

Purpose

A focused independent verdict for Go / No-Go / Re-Architect decisions where a SAP BTP extension or AI architecture decision cannot converge.

Best for

  • Urgent decision impasse.
  • Protected-path failure such as AppRouter / Work Zone divergence.
  • First signs of tenant rollout or lifecycle divergence.
  • Executive-level need to classify whether the issue is structural.

Deliverable

Boundary Verdict Memo with P1 structural risks, violated boundaries, and a Go / No-Go / Re-Architect direction. Typical duration: 5-7 business days.

Request boundary verdict
Audit

Full Boundary Audit

Purpose

A deeper architecture audit for SAP BTP landscapes that must survive long-term evolution, scale, governance, security hardening, and multi-tenant rollout.

Best for

Recurring incidents, fragmented ownership, vendor disagreement, rollout instability, or a landscape that must survive tenant, identity, data, integration, and lifecycle evolution.

Deliverable

Boundary risk matrix, failure-mode analysis, ownership/jurisdiction map, and final architectural verdict with non-implementation remediation direction.

Typical duration: 2-3 calendar weeks.

Request audit scope
Enterprise AI

Agent-Extension Responsibility Review

Purpose

To determine whether an SAP Joule, enterprise AI, workflow, or agent initiative has clearly assigned responsibility across agents, governed tools, extension APIs, workflows, assistants, and core systems.

Typical questions

  • Should this be an agent, an extension, a workflow, or automation?
  • What business truth is involved?
  • What can the agent execute?
  • Which workflow contains accountability?
  • Which extension APIs or governed tools should be invoked?
  • Which system owns final state?
  • How will prompts, tools, policies, workflows, context, and extension APIs evolve?

Deliverable

Responsibility Boundary Memo with a responsibility map, ambiguity areas, truth / execution / lifecycle / accountability risks, and recommended architecture direction.

Assess responsibility boundaries
Authority

Architecture Verdict Authority

Purpose

Programs where a single architectural misjudgment would create disproportionate long-term cost and recurring external arbitration is needed.

Deliverable

Periodic boundary health reviews, pre-validation of major architectural changes, and limited emergency architectural arbitration.

Send structured summary

Why this is different from implementation review

Implementation reviews validate local correctness; boundary work establishes whether the architecture is still legitimate across responsibility, runtime, identity, tenant, data, integration, and lifecycle boundaries.

The question is not whether a capable team can keep the initiative moving. The question is whether the current structure deserves to continue as designed.

This preserves independence. Delivery teams implement; this service judges whether the architecture deserves to proceed.

Out of scope

  • Not implementation work.
  • Not configuration work.
  • Not production code maintenance.
  • Not staff augmentation.
  • Not project management.
  • Not delivery team replacement.

Initiation protocol

Send one structured page. That is enough to determine whether the work is an agent-extension responsibility review, strategic verdict, full audit, or verdict authority format.

  • Landscape — S/4, BTP runtimes, entry surfaces, identity
  • Divergence — by tenant, path, surface, agent behavior, or workflow state
  • Timeline — go-live, rollout, recent change, AI pilot, model/tool change
  • Ownership — who holds truth, action, state, lifecycle, and accountability today
  • Decision — the choice that cannot converge
Send structured summary Use intake template

Related reading and SEO entry pages

These pages explain common situations and architecture models. The actual engagement format is determined after the one-page summary, usually as a Strategic Boundary Verdict, Full Boundary Audit, Agent-Extension Responsibility Review, or Architecture Verdict Authority path.