Skip to content
GREEN FUTURETECH.
Explore
The illustrative default portfolio on the maturity and return planeSix programme use cases. Bubble area is data need. 1 case, computer-vision defect detection, passes the phase and data gates with a priority of 76.5; 5 are gated. All scores are illustrative.T3 EXPLORET2 NEXTT1FIRST005050100100MATURITY →RETURN OPPORTUNITY →Computer-vision defect detection: maturity 90, return 85, data need 20. Eligible.01ML demand forecasting: maturity 85, return 80, data need 60. Gated.02Inventory and supply-chain optimisation: maturity 70, return 75, data need 60. Gated.03Digital printing and sample reduction: maturity 65, return 60, data need 45. Gated.04Multi-model design-to-spec pipelines: maturity 45, return 65, data need 80. Gated.05Physics-constrained generation and GenAI content: maturity 30, return 50, data need 90. Gated.06MEASURING WEFT

76.5priority, case 01
out of 100, illustrative

DEFAULT / ILLUSTRATIVE · 1 ELIGIBLE · 5 GATED · 6 CASES

GFT instruments / Digital decisions

Start with readiness.
Then follow the return.

Compare six apparel use cases by maturity, return opportunity and data need. Test the tier, then check the phase and data gates. Every numeric score is illustrative.

Sequence the portfolio
Operate the instrumentProcedural visualisation. Schematic, not a facility.

An instrument for the decision

Which machine-learning use cases first?

Vision leads the illustrative launch case. Forecasting stays a first-tier priority but waits for its data and scale phase. Change any input: the plane re-seats every disc and re-rings the gates.

Scenario

Presets keep the programme’s six use cases. Scores, capacity and tier thresholds are illustrative planning assumptions, not GFT operating results.

Operating phase

A screening proxy for relevant history, labels, access and quality. It is not a count of records.

Selected case

Programme Tier 1. Scores below are illustrative.

Is this approach proven in your operating context?

How material is the addressable problem?

How demanding is the required data and labelling?

A dependency floor. The scenario tier can postpone it further.

Add / compare

New cases start at illustrative scores of 50 for maturity, opportunity and data need, with a Phase 2 floor. Maximum 12 cases.

Nothing is saved to a server. A is held in this page until you share its link.

Save scenario A, then edit B. A stays on the plane as dashed ghosts.

Computer-vision defect detection
76.5priority points
out of 100

Tier 1 · Eligible for investigation

1eligible

5gated

6cases

1 eligible + 5 gated = 6 cases

Computer-vision defect detection leads the eligible set in Tier 1. 5 of 6 cases stay gated: confirm a baseline and run a bounded pilot first.

Phase and data gates pass. This is not deployment approval.

Maturity against return opportunity, bubble area is data needIllustrative scores from 0 to 100. 1 of 6 use cases pass the phase and data gates. Tier 1 is the upper right zone, Tier 2 the middle zone and Tier 3 the rest. Solid rings pass; dashed rings are gated. Exact values are in the ledger table. T3 EXPLORET2 NEXTT1FIRST 005050100100 MATURITY →RETURN OPPORTUNITY → Computer-vision defect detection. Tier 1. Maturity 90, return opportunity 85, data need 20. Eligible for investigation.01ML demand forecasting. Tier 1. Maturity 85, return opportunity 80, data need 60. Needs Phase 2 (Scale). Data gap 20 points.02Inventory and supply-chain optimisation. Tier 2. Maturity 70, return opportunity 75, data need 60. Needs Phase 2 (Scale). Data gap 20 points.03Digital printing and sample reduction. Tier 2. Maturity 65, return opportunity 60, data need 45. Needs Phase 2 (Scale). Data gap 5 points.04Multi-model design-to-spec pipelines. Tier 3. Maturity 45, return opportunity 65, data need 80. Needs Phase 3 (Optimise). Data gap 40 points.05Physics-constrained generation and GenAI content. Tier 3. Maturity 30, return opportunity 50, data need 90. Needs Phase 3 (Optimise). Data gap 50 points.06 P0 SET UPP1 LAUNCHP2 SCALEP3 OPTIMISE Computer-vision defect detection: earliest Phase 1. Eligible01ML demand forecasting: earliest Phase 2. Needs Phase 2 (Scale). Data gap 20 points02Inventory and supply-chain optimisation: earliest Phase 2. Needs Phase 2 (Scale). Data gap 20 points03Digital printing and sample reduction: earliest Phase 2. Needs Phase 2 (Scale). Data gap 5 points04Multi-model design-to-spec pipelines: earliest Phase 3. Needs Phase 3 (Optimise). Data gap 40 points05Physics-constrained generation and GenAI content: earliest Phase 3. Needs Phase 3 (Optimise). Data gap 50 points06

Exact values and their sources

Read every input beside its result.

All scores are illustrative points and tiers are ordered categories. Point at a row and the lens finds its disc; choose a source layer and the table shows what that layer feeds.

Status lens

2 sourced 4 illustrative

Six default inputs Status Point at a segment, or choose a layer, to light what it feeds.

Names and tiers Section 3 Six cases, two per tier, from the digital strategy, section 3.

Scores Illustrative Engineering judgement. No measured data supports maturity, opportunity or data need.

Phase and capacity Illustrative Phase 1 and capacity 40: an illustrative launch scenario, not a count of records.

Dependency floors Section 7 Vision 1; forecasting and inventory 2; generative applications 3, from section 7. Printing’s floor is illustrative.

Presets Illustrative Scale: phase 2, capacity 75. Mature: phase 3, capacity 100. Scores stay unchanged.

Thresholds Illustrative Tier cut-offs and the priority equation were made for this instrument. Not an industry standard.

Show

Sorted by case order.

Current scenario, all use cases. S3 and S7: sourced, digital strategy sections 3 and 7.
Prog. tierFloorReadinessStatus
01Computer-vision defect detection9085201P1176.5EligibleS3S7Scores
02ML demand forecasting8580601P2168Needs P2 · gap 20S3S7Scores
03Inventory and supply-chain optimisation7075602P2252.5Needs P2 · gap 20S3S7Scores
04Digital printing and sample reduction6560452P2 *239Needs P2 · gap 5S3Floor *Scores
05Multi-model design-to-spec pipelines4565803P3329.3Needs P3 · gap 40S3S7Scores
06Physics-constrained generation and GenAI content3050903P3315Needs P3 · gap 50S3S7Scores

GFT’s Cameroon textile programme, digital strategy, sections 3 and 7 (prepared 31 January 2024, sources refreshed 2026). Use-case structure only; all numeric scores and screening thresholds are illustrative GFT assumptions. Not offered for download here. The asterisk marks the illustrative Phase 2 floor for digital printing.

The method, ordinal not financial

A first-tier idea can still need to wait.

The screening method in five poses of one planeProgramme tiers as three columns; the scored plane; the priority rectangle for the selected case with curves of equal priority; the two tier cut-offs; then the plane tilted into a floor. All scores are illustrative. PROGRAMME TIER 1PROGRAMME TIER 2PROGRAMME TIER 3Vision defect detectionDemand forecastingInventory optimisationDigital printingDesign-to-spec pipelinesPhysics-constrained generation 005050100100MATURITY M →RETURN R →AREA = DATA NEED P25P50P75 T1T2T3 M 80 / R 70M 55 / R 50 010203040506 80 × 70 / 100 = 56
Data capacity wall at 40: each case's data need, with its gap above the wall050100DATA CAPACITY 40+5006+4005+504+2003+2002PASS01DATA NEED BY CASE, ORDERED BY MATURITY

Two cases per programme tier, from the digital strategy, section 3. These are the programme’s priorities, rather than measured results.

A return score is not a percentage saving, a rate of return, a price or a forecast of cash flow. Area is data need.

90 × 85 / 100 = 76.5 · 76.5 % of the plane. Curves join equal priority: 25, 50, 75 (illustrative).

Cases now: 2 / 2 / 2 in T1 / T2 / T3. The cut-offs are illustrative design choices, not an industry standard.

At a capacity of 40: 1 of 6 cases pass under the wall. Gaps: 20, 20, 5, 40, 50.

  1. 01 / 05

    The programme sets the order.

    The programme puts computer-vision defect detection and machine-learning demand forecasting in its first tier. Inventory optimisation and digital printing follow in the second. Multi-model design pipelines and physics-constrained generation sit in the emerging third tier.

    Sourced: digital strategy, section 3

  2. 02 / 05

    Score every case on one boundary.

    This instrument adds an explicit, illustrative scoring scheme so that a team can challenge that structure. Maturity is confidence in the approach for the intended process. Return opportunity is the materiality of the problem. Data need is the burden of usable history, labels and access. Neither score is a probability.

    Illustrative scoring scheme

  3. 03 / 05

    Priority is a product.

    Maturity and opportunity each run from 0 to 100, so priority is the share of the plane that a case’s rectangle covers. Maturity 80 and opportunity 70 give 56 points.

    P = M × R / 100

    Selected case: 90 × 85 / 100 = 76.5

    Illustrative equation

  4. 04 / 05

    Tiers are cut-offs.

    Tier 1 needs maturity of at least 80 and opportunity of at least 70. Otherwise Tier 2 needs at least 55 and 50; the rest fall into Tier 3, and equality passes. The cut-offs are illustrative choices that reproduce the programme’s starting order.

    T1: M ≥ 80, R ≥ 70Illustrative thresholds

  5. 05 / 05

    Readiness is a separate gate.

    The priority index orders cases within a tier. It cannot remove a dependency. The data gap is max(0, data need minus capacity): the ochre part of each bar above the wall.

    phase ≥ max(tier, floor) · capacity ≥ data need · opportunity > 0Screening gate, not deployment approval

Before anyone scores

A beautiful plot cannot compensate for inconsistent scoring.

  1. 01 Computer-vision defect detection, maturity 90
  2. 02 ML demand forecasting, maturity 85
  3. 03 Inventory and supply-chain optimisation, maturity 70
  4. 04 Digital printing and sample reduction, maturity 65
  5. 05 Multi-model design-to-spec pipelines, maturity 45
  6. 06 Physics-constrained generation and GenAI content, maturity 30

0 means: never run at this station100 means: proven at this station

0 means: untried on this product family100 means: proven across the family

0 means: no evidence for this decision100 means: proven for this decision

Score each case against the same operating boundary. Compare applications only when the team agrees what the low and high ends mean. Record a short rationale outside this tool.

Choose the boundary

One inspection station. Write down what a maturity of 0 and of 100 means here, and what a return of 0 and of 100 means here, before the first score.

One product family. Write down what a maturity of 0 and of 100 means here, and what a return of 0 and of 100 means here, before the first score.

One planning decision. Write down what a maturity of 0 and of 100 means here, and what a return of 0 and of 100 means here, before the first score.

Read WP02 Measure Before You Model A title from the GFT paper programme; no released PDF is claimed. The bracket width is illustrative.

Source boundary

The programme supplies the sequence, not a forecast.

The programme sequence begins with capture and bought vision, adds forecasting once sales history exists and reserves emerging generative applications for later pilots.

P0 / SET UP

Set up opens no use case.

In the default scenario no case has an earliest phase of 0, so this phase builds the data spine only. Every use case waits on its thread.

WP14 The Industrial Data Spine

P1 / LAUNCH

Launch opens the first-tier vision case.

Computer-vision defect detection is the only case whose earliest phase is 1. Its floor comes from the digital strategy, section 7, and its illustrative data need sits under the default capacity of 40, so it drops through the weft.

Sourced floor, illustrative scores

P2 / SCALE

Scale adds forecasting, inventory and printing.

Forecasting and inventory carry sourced Phase 2 floors; digital printing uses an illustrative one. All three need more data capacity than the default of 40, so they stop at the weft. Raise capacity in the instrument to open them.

Sourced floors, illustrative printing *

P3 / OPTIMISE

Optimise is the earliest phase for the emerging cases.

Multi-model design pipelines and physics-constrained generation sit in the programme’s third tier with Phase 3 floors. Their data needs are the highest in the set, so they hang highest above the weft and wait longest.

Sourced floors, illustrative scores

Reading P1 Launch · 1 of 6 cases pass phase and data gates at a capacity of 40.

Four phase lanes and the data-capacity weftEach case hangs on a thread above its earliest phase. When that phase is read, a case whose data need the capacity covers drops through the weft; the others stop above it by their gap. Capacity 40. Scores are illustrative. P0 SET UPP1 LAUNCHP2 SCALEP3 OPTIMISE DATA SPINE THROUGH THE GATES ↓ DATA CAPACITY 40 01+002+2003+2004+505+4006+50
Move the instrument’s phase

What the gates leave out

Use the shortlist to design a test.

Passing the gates does not demonstrate accuracy, security, rights to use data, safe process integration or economic value. A scalar capacity score cannot tell sales history from defect labels. Design the test before the pilot.

Design the test for

Inspect real fabric varieties and difficult defects: 3 of 18 swatches carry a hard defect in this illustrative layout.

Sector literature, context only, not used to calibrate any score.

Read WP03 Seeing Every Defect

Connect priority to measured loss: Operate the OEE loss instrument ↗

Back-test against a simple baseline: the last 40 percent of the series is held out, with no future information in training.

Sector literature, context only, not used to calibrate any score.

Read WP04 Forecasting From a Cold Start

Establish a production-loss baseline before scoring a remedy.

Check that generated output becomes a manufacturable specification: 3 hand-overs, each with human review and a rollback route.

No external source is cited for this test.

Read WP14 The Industrial Data Spine

Paper titles from the GFT programme; no released PDF is claimed.

Not modelled: capital cost, operating cost, implementation capacity, vendor dependency, downtime, energy, financing. Build a separate costed business case. Not investment, legal or tax advice.

The selected case at lens scale, with the test drawn inside18 SWATCHES, 3 HARD DEFECTSREAL FABRIC VARIETIESTRAIN: HISTORYTEST, LAST 40%SIMPLE BASELINEGENERATEDSPECMAKEABLE?REVIEW AND ROLLBACK AT 3 HAND-OVERSCASE 01 AT LENS SCALE · ILLUSTRATIVE
The gate dialFour phase sectors, P0 at twelve o’clock. Each case sits in its earliest phase sector; cases that pass at the chosen phase and capacity move in to the hub ring. The hub holds the shortlisted case; four notches mark the four pilot checks. P0 SET UPP1 LAUNCHP2 SCALEP3 OPTIMISE 010203040506 01ELIGIBLE

Use the result in its boundary

Score it once, sequence it, then pilot it.

Use the shortlist to assign a problem owner, a baseline and an acceptance test. A high score is a reason to investigate, not evidence that an application will work in your factory.

Turn the dial to a phase
Tick each before a pilot is scoped.

Nothing settles in P0 Set up. The data spine comes first and every case waits.

Computer-vision defect detection settles in P1 Launch, the earliest phase its sourced floor allows.

P2 Scale adds forecasting, inventory and printing beside vision. Each still needs the data capacity.

P3 Optimise is where the emerging cases settle, once the earlier phases and data are in place.

Investigate 0 of 4 checks. Complete the checklist to scope a pilot. This is not deployment approval.

Return to the instrument ↑Read the measurement approach ↗

Find a thread.

Type to search.