# Northstar SCALE Walkthrough

**SCALE version:** 0.5  
**Walkthrough revision:** Corrected, 2026-08-22  
**Status:** Completed fictional example  
**Current input:** packaged file `examples/NORTHSTAR-SOURCE-CASE.md`, a canonical copy of `northstar-equipment-services-case-v1.1.md`; **Episode A only**

## Scope of this corrected walkthrough

The original SCALE v0.5 package used the Northstar v1.0 source case. This corrected walkthrough uses the canonical identifiers and Episode A content in Northstar v1.1. It intentionally does not use Episode B readiness evidence or Episode C results to make the Episode A recommendation appear more certain than it was.

This walkthrough shows how the portable SCALE agent can turn one supplied case into a bounded architecture analysis. It is an example of method and output structure, not a source of enterprise facts for another analysis.

## 1. Launch and charter

The architect loads the kit, attaches the fictional Northstar source case, pastes the launch prompt, and asks:

> Which bounded response, if any, should Northstar fund now to improve service-promise reliability, preserve usable branch discretion, protect margin, reduce avoidable customer disruption and repeat work, and learn whether enterprise scheduling technology is necessary?

SCALE proposes the minimum charter:

| Field | Recorded value |
|---|---|
| Decision supported | Select, condition, defer, or reject alternatives A0–A3; keep the later platform decision distinct |
| Decision owner | ORG-13 Investment Council |
| Sponsor | ORG-09 EVP Service Operations |
| Primary analyst | ORG-14 Business architect |
| Included scope | O-01/O-02/O-03, M-01–M-08, service-promise policy, required part/skill/capacity signals, relevant capabilities and objects, three-branch change, evidence, cost/capacity bands, platform dependency |
| Excluded scope | ERP replacement, R2/R3 authorization, enterprise rollout, vendor selection, detailed solution design, committed delivery estimate |
| Evidence cutoff | Episode A: E-001 through E-007 |
| Required work product | Architecture-backed investment recommendation with alternatives, conditions, evidence, uncertainty, and reopen triggers |
| Sufficiency condition | The Investment Council can distinguish the business change worth testing now from technology procurement that remains unproven |
| **Analysis stop condition** | Stop analysis when the forum can compare A0–A3, understand the material architecture path and uncertainty, assign unresolved conditions, and take or defer the decision; additional modeling would not change that choice |
| Reopen trigger | New evidence, a changed cost/capacity assumption, unresolved ownership, or a proposed scope change could alter the recommendation |

**Release-entry blockers are not the analysis stop condition.** An unapproved promise definition, unresolved BO-05 ownership/source status, inadequate sampling, or no credible measurement method becomes an explicit condition preventing R1 readiness. It does not justify endless modeling or allow SCALE to decide on behalf of the accountable forum.

**Human disposition:** Accept.

## 2. Source map and evidence register

SCALE uses `SRC-*` only as location references into the supplied Markdown. A source section is not itself evidence. Evidence retains the case's canonical `E-*` identifier wherever it is proposed, accepted, reused, or cited.

### Source-section map

| Source ID | Source location | Use in this walkthrough |
|---|---|---|
| SRC-001 | Enterprise context | Establish fictional enterprise boundary |
| SRC-002 | Executive request and architecture decision | Establish the decision, sponsor, and authority |
| SRC-003 | Strategic intent and canonical measure register | Establish O-01/O-02/O-03, M-01–M-08, targets, and provisional definitions |
| SRC-004 | Stakeholders, actors, and decision rights | Establish ORG-* contributions and rights |
| SRC-005 | Value stream | Establish VS-01 and ST-01–ST-05 |
| SRC-006 | Capabilities and clean boundaries | Establish C-100–C-180 and their boundaries |
| SRC-007 | Business objects and lifecycles | Establish BO-01–BO-06 and unresolved source status |
| SRC-008 | Evidence register and lineage | Locate canonical evidence records |
| SRC-009 | Alternatives, release logic, sampling, and guardrails | Establish A0–A3 and R1 boundaries |
| SRC-010 | Episode A decisions and open items | Establish DEC-001, DEC-002, and OI-01–OI-05 |

### Canonical Episode A evidence

| Evidence ID | Source location | Claim supported | Confidence / authority | Limitation |
|---|---|---|---|---|
| E-001 | SRC-008 | M-01 is approximately 82%; promise failures and customer escalations are material | High directionally | Regional definitions and cancellation handling differ |
| E-002 | SRC-008 | M-02 is approximately 14% | Medium | Sample selection and cause coding are disputed |
| E-003 | SRC-008 | Promises often precede part, skill, and capacity confirmation; exception practices vary | Medium-high as convergent observation | Workshops and observation are not a controlled measure |
| E-004 | SRC-008 | The proposed platform can automate several desired controls | Medium for product functionality | Does not establish enterprise fit, adoption, or realized value |
| E-005 | SRC-008 | M-03 gross service margin is 31% | High | Enterprise result may mask branch and work-type variation |
| E-006 | SRC-008 | M-04 is approximately 38% under the provisional definition | Medium | Two-week purposive sample; definition not approved |
| E-007 | SRC-008 | Candidate branches expose different operating conditions and bounded capacity | Medium | Planning evidence, not proof of adoption or representativeness |

SCALE does not create a second evidence namespace for these records. A downstream work product cites `E-001`, not a tool-specific duplicate.

## 3. Proposed architecture

SCALE proposes only the architecture required for the Episode A investment decision.

### Elements

| ID | Type | Name | Status before review | Source / evidence |
|---|---|---|---|---|
| O-01 | Objective | Improve promise reliability with appropriate branch discretion | Proposed | SRC-003; E-001, E-003 |
| O-02 | Objective | Protect service margin | Proposed | SRC-003; E-005, E-007 |
| O-03 | Objective | Reduce avoidable customer disruption and repeat work | Proposed | SRC-003; E-001, E-002, E-003 |
| VS-01 | Value stream | Restore Equipment Availability | Proposed | SRC-005 |
| ST-02 | Value-stream stage | Define service promise | Proposed | SRC-005 |
| C-110 | Capability | Service Promise Management | Proposed | SRC-006; E-003 |
| C-130 | Capability | Resource Scheduling | Proposed | SRC-006; E-003, E-007 |
| C-140 | Capability | Parts Availability Management | Proposed | SRC-006; E-003 |
| C-170 | Capability | Service Performance Management | Proposed | SRC-006; E-001, E-002, E-005, E-006 |
| C-180 | Capability | Business Information Governance | Proposed | SRC-006, SRC-007 |
| BO-02 | Business object | Customer Promise | Proposed | SRC-007 |
| BO-05 | Business object | Part Reservation | Proposed, owner/source unresolved | SRC-007; E-003 |
| BO-06 | Business object | Technician Assignment | Proposed | SRC-007; E-003 |
| I-01 | Initiative | Connected Service Promise | Proposed | SRC-009 |
| R1 | Release | Three-branch feasibility release | Proposed | SRC-009; E-007 |

### Relationships

| ID | Proposed relationship | Evidence / source | Initial confidence |
|---|---|---|---|
| RL-001 | O-01, O-02, and O-03 frame the outcome and guardrails for VS-01 | SRC-003, SRC-005 | High as case architecture |
| RL-002 | VS-01 contains ST-02 | SRC-005 | High as case architecture |
| RL-003 | C-110 enables ST-02 and uses signals from C-130 and C-140 | SRC-005, SRC-006; E-003 | Medium-high |
| RL-004 | C-140 produces BO-05 status used by C-110; the authoritative source remains unresolved | SRC-006, SRC-007; E-003 | Medium |
| RL-005 | C-170 governs M-01–M-08, while named business and Finance authorities remain accountable for definitions and outcomes | SRC-003, SRC-004, SRC-006 | High as role design; provisional operationally |
| RL-006 | R1 changes C-110, C-130, C-140, C-170, and C-180 without replacing ERP | SRC-009 | Medium-high as proposed design |
| RL-007 | A3 platform procurement should depend on repeatability evidence and residual automation gaps, not vendor functionality alone | E-004; SRC-009 | High as recommendation logic |

## 4. Proposal review

SCALE asks for human disposition rather than treating extraction as acceptance.

| Proposal | Disposition | Rationale |
|---|---|---|
| O-01/O-02/O-03, VS-01, ST-02, C-110, C-130, C-140, C-170, C-180, BO-02, BO-06, I-01, and R1 | Accept for this fictional engagement | Definitions and scope are explicit in the supplied case |
| ORG-09 is accountable for C-110/C-170 performance; architecture stewards maintain their records | Accept | Preserves the distinction between performance accountability and architecture stewardship |
| ORG-08 is the BO-05 owner and the inventory reservation record is authoritative | Investigate | Both are proposed in Episode A and require OI-02 disposition |
| Purchasing the proposed platform will produce O-01/O-02/O-03 value | Reject | E-004 supports product functionality, not enterprise fit, adoption, causation, or realized value |
| A1/R1 can test the highest-risk business assumptions without ERP replacement | Accept as a recommendation | Supported by the bounded release design; actual feasibility remains unknown at Episode A |
| R1 results will be representative of all 48 branches | Reject | E-007 supports purposeful variation, not statistical or enterprise representativeness |

## 5. Trace and analysis

### Direct finding

Northstar should conditionally fund **A1/R1**, within the provisional $150,000–$300,000 planning band, and defer **A3** enterprise platform procurement.

This is a recommendation, not a prediction that R1 will succeed. A0 does not materially test the operating problem. A2 commits regional capacity before the basic policy and information model are proven. A3 creates high cost and path dependence while the causal problem and residual automation need remain uncertain.

### Architecture path

`O-01/O-02/O-03 → VS-01/ST-02 → C-110 + C-130 + C-140 + C-170 + C-180 → BO-02/BO-05/BO-06 → I-01/R1 → M-01–M-08 → ORG-13 decision`

### Principal evidence

E-001 through E-007, with particular reliance on E-003 for the observed operating problem, E-005 for the margin guardrail, E-006 for the provisional confirmation baseline, and E-007 for the bounded branch design.

### Conflicts and gaps

- M-01 definitions and cancellation handling differ by region.
- M-02 sampling and cause coding are disputed.
- M-04–M-08 include provisional definitions, baselines, or tolerances.
- BO-05 ownership and authoritative-source status remain unresolved.
- Branch exception authority is not yet governed through POL-01.
- E-004 proves product functionality only.
- Cost and capacity bands are sufficient for comparison but not a committed estimate.

### Confidence

**Medium-high in the recommendation to test a bounded operating change; low in any claim about scale or vendor value.** The problem, architecture boundary, alternatives, and reversibility are explicit. Confidence is reduced by provisional measures, purposive sampling, unresolved information authority, and the absence of Episode B/C evidence at this decision date.

### Remaining human judgment

The Investment Council must decide whether the bounded evidence gain justifies the cost and opportunity cost. Operations, Customer, Supply, Information Governance, and Finance must approve their definitions, rights, and guardrails. The business architect and SCALE cannot make those decisions.

## 6. Work product

SCALE prepares an architecture-backed investment brief containing:

1. the direct A0–A3 recommendation and planning bands;
2. the O/M/VS/ST/C/BO/ORG/I/R architecture path;
3. R1 scope, purposeful branch criteria, and exclusions;
4. E-001–E-007 lineage, confidence, conflict, and missing evidence;
5. the distinction between accountable executives, owners, and stewards;
6. conditions that must be resolved before a readiness decision;
7. governed guardrails and breach authorities;
8. the separate R2 repeatability and platform-procurement decisions; and
9. analysis stopping and reopening rules.

## 7. Recorded Episode A human decisions

The v1.1 case supplies exactly two initial governed decisions:

- **DEC-001:** Conditionally authorize A1/R1 within the planning band, subject to approved POL-01 and measure definitions, confirmed object owners and source statuses, purposeful branch selection, a readiness decision, and 30/90-day review gates.
- **DEC-002:** Defer A3 enterprise platform procurement. Reopen only after R2-quality repeatability evidence identifies material residual automation gaps and a separate business case compares vendor and non-vendor responses.

SCALE records the decisions, conditions, owners, and triggers. It does not claim to have made them. Episode B and C decisions are out of scope for this walkthrough because they were not available at Episode A.

## 8. Continuing architecture

The architect exports accepted elements, relationships, evidence, decisions, and open items without changing their canonical identifiers. A later session can ask:

> Is Northstar's R1 operating model ready to begin under the conditions in DEC-001?

That session may use Episode B and E-008. A subsequent 30- or 90-day session may use E-009 or E-010 only after those records exist. It reuses the accepted architecture and adds actual readiness, adoption, segmented performance, guardrail, and exception evidence. Architecture accumulates without rewriting the earlier recommendation through hindsight.

## 9. What this example demonstrates

The portable kit can guide a repeatable path from a business question through source mapping, canonical evidence, proposed architecture, human disposition, traceable analysis, a decision work product, a decision record, and continuing reuse. It does not prove that an AI answer is correct, that the fictional recommendation applies elsewhere, that Episode A establishes readiness or value, or that a general AI platform supplies enterprise governance controls.