The research, led by Chirag Thummar, tackles a problem that has long frustrated mid-sized manufacturers operating between 10 and 50 presses. At this scale, the study argues, the dominant source of waste is not poor molding engineering but decision latency: the gap between when a decision is needed and when it is actually made with adequate information. Slow quotes, fuzzy scheduling, and unranked maintenance priorities force managers to compensate with excess inventory, overtime, expedited freight, and redundant inspection—each buffer adding cost and instability across the value stream.

Lean manufacturing tools such as 5S, standardized work, value stream mapping, Kanban, SMED, and total productive maintenance have long delivered gains in molding environments, including changeover reductions of 39 to 54 percent reported in prior studies. But in high-mix plants, lean implementations tend to plateau once decision complexity exceeds what standard work and visual management can absorb. The new study’s central claim is that this is precisely where AI belongs—not as an autonomous optimizer, but as a constrained decision accelerator operating inside a lean-defined architecture.

The system was tested at a facility running 33 presses between 55 and 720 tons of clamp force, supplying aerospace, automotive, and medical customers under ISO-certified quality systems. The plant juggled roughly 450 active stock-keeping units, frequent mold changes, and volatile order volumes across two to three shifts. At baseline, quote turnaround averaged 48 hours, schedule adherence sat at 78 percent, overall equipment effectiveness was 58 percent, and work-in-process inventory stood at 12 days.

The architecture has three layers. The first, a lean foundation, establishes operational stability through 5S, standardized work at each press, visual controls, and daily tiered management routines—while simultaneously creating a disciplined data-capturing environment. The second, data enablement, builds lightweight pipelines from existing ERP and MRP exports, downtime logs, maintenance tickets, and quality records, cleaning and time-stamping everything without sensors or IoT devices. The third layer provides interpretable AI outputs in three domains: quote generation via similarity-based retrieval across part family, material, and complexity; finite-capacity scheduling with setup-aware dispatch; and loss prioritization through Pareto-ranked downtime, scrap, and delivery signals.

Notably, the AI modules are rule-based and weighted rather than machine-learning models, a choice the author defends on three grounds. The facility’s roughly 450 SKUs and inconsistently coded records could not satisfy the data-to-feature ratios that gradient-boosted trees or sequence models demand without overfitting. Interpretability was a hard governance requirement, because every AI recommendation had to be accepted or overridden by a human at two formal gates, and overrides must be traceable to specific features. And when upstream lean layers already shrink the decision space, the marginal value of algorithmic sophistication drops sharply—a view consistent with prior work on the cultural tensions of AI adoption in lean quality management.

Two governance gates controlled the flow of orders. Gate 1, Quote Review, required documented routing assumptions, cycle-time estimates, and tooling requirements before any quote was released, with the AI supplying an analog-based recommendation that a quote engineer confirmed or overrode. Gate 2, Schedule Release, verified data completeness, downtime coding consistency, training readiness, and system validation before an order reached a press, with the scheduler explicitly confirming or overriding the AI’s ranked dispatch advice. Overrides were logged and used to recalibrate the model weights monthly, while a feedback loop routed scrap, downtime, and delivery signals back into daily improvement routines.

The results, drawn from 25 paired production lines each with a distinct press-mold-part-family combination, were striking. Cycle time fell from 47.95 seconds to 34.68 seconds, a 27.7 percent reduction. Scrap dropped from 6.46 percent to 2.11 percent, bringing the facility below its 3 percent internal target. Unplanned downtime fell from 2.65 hours per shift to 1.37 hours, and changeover time dropped from 78.48 minutes to 46.36 minutes. All four improvements were statistically significant after Bonferroni correction, with Cohen’s d effect sizes between 0.85 and 6.07—every one exceeding the conventional large-effect threshold. The scrap effect was so large that the author ran sensitivity checks: excluding the most extreme lines still yielded effect sizes above 5, and a decomposition of nearly 11,000 scrapped parts showed that transient-state events such as startups and changeovers accounted for 67.4 percent of scrap and fell by 56.6 percent during deployment, consistent with the standardized startup checklists and first-piece approval gates introduced under the lean foundation.

Principal component analysis reinforced the picture of a system-level shift rather than isolated gains. The first two components explained 78.7 percent of the variance, and pre- and post-implementation observations separated cleanly along the first axis, with post-implementation points clustering tightly. The correlation structure itself changed: the pre-implementation negative correlation between cycle time and downtime—evidence of frantic, compensatory schedule recovery—weakened from −0.51 to −0.09, suggesting that AI-supported dispatching absorbed disruptions at the scheduling level instead. Process variability fell by 15 to 26 percent across all four key performance indicators, and a target-based capability index for scrap improved from a negative value, meaning the process average sat above the internal target, to 1.24, indicating genuine capability.

Sustainability was tracked over 26 weeks after the deployment ended, and the gains held. Overall equipment effectiveness climbed steadily from 65.6 percent in week one to 72.0 percent by week 26, scrap converged to about 2.2 percent, and schedule adherence stayed within a narrow band. All four stability indices remained below the facility’s 0.006 target for mature, well-governed processes. Downtime coding accuracy, a proxy for data quality, improved from 88.2 percent in month one to 96.2 percent by month 12, creating what the author describes as a virtuous cycle in which lean discipline strengthens the data infrastructure, which in turn reinforces the reliability of AI recommendations. The line-level cost-benefit analysis yielded annualized savings of USD 563,250 under a full-impact scenario and USD 487,250 under a conservative estimate counting only directly metered categories, against an implementation cost of roughly USD 15,000.

The study is candid about its limits. It is a single-facility investigation, so the results cannot be extrapolated without multi-site replication, and the overlapping deployment phases prevent clean causal isolation of lean versus AI contributions. Facility-level trajectories show no distinct acceleration after AI activation, suggesting the lean foundation drove much of the early improvement. Still, the practical message for the thousands of mid-sized molding plants worldwide is clear: deploy lean as a system-level architecture first, constrain AI within governance gates as decision input rather than authority, and recognize that existing ERP data combined with lean-disciplined capture is enough—sensors and IoT are not prerequisites. Future work will pursue multi-site replication, closed-loop scheduling, and longer tracking windows to confirm that these durable gains generalize.

Subject of Research: Lean-AI integration for waste reduction and cost optimization in medium-scale injection molding

Article Title: A lean–AI integration for decision-centric waste reduction and operational cost optimization in medium-scale injection molding

Article References: Thummar, C. (2026). A lean–AI integration for decision-centric waste reduction and operational cost optimization in medium-scale injection molding. Heliyon, 12(15), Article e45517. https://doi.org/10.1016/j.heliyon.2026.e45517

Image Credits: AI Generated

DOI: 10.1016/j.heliyon.2026.e45517

Keywords: injection molding, lean manufacturing, artificial intelligence, waste reduction, governance gates, SMED, overall equipment effectiveness, scrap reduction, decision latency, ERP data, process capability, cost-benefit analysis

The scientific foundation of DQPA rests on two pillars of international standardization. ISO 8000 establishes the principles of data quality management, while the ISO/IEC 33000 family provides the general mechanism for process capability assessment: a process reference model, process attributes rated on an ordinal scale, and a maturity model that aggregates those attributes into organizational levels. Although this mechanism has a long track record in domains such as software development, Green IT, and data quality certification, no international instantiation had ever covered data governance, data management, and data quality management jointly as integrated disciplines. The only ISO instantiation for the data domain, the ISO 8000-6x series, is confined to data quality management alone. The researchers built on a direct precedent, the MAMD model, and on the UNE 0077 through 0080 specifications—what they describe as the first standardization initiative to close that gap with normative status—enriched with the governance principles of ISO/IEC 38505 and the DAMA-DMBOK body of knowledge.

What distinguishes self-assessment from formal certification is its purpose. Under ISO/IEC 33000, assessment by an independent team enables certification with validity toward third parties, while self-assessment, performed by the organization itself, is an equally recognized application of the same method aimed at understanding one’s own situation recurrently and affordably as a basis for continuous improvement. Until now, no adequate instrument has existed for this second application, for two reasons. Independent assessment is too costly to repeat frequently, and its most expensive phase—evidence collection—depends heavily on tacit human knowledge that is difficult to automate. Moreover, translating normative processes into language that business profiles can act upon, one of the very purposes of data governance, rarely occurs in manual practice. Existing frameworks fall short in different ways: COBIT 2019 lacks an integrated data-domain maturity model; DCAM and CMMI-DMM support self-assessment but are not grounded in ISO/IEC 33000; and DAMA-DMBOK systematizes the disciplines without defining a maturity model of its own. None combines integrated coverage, an ISO/IEC 33000-based mechanism, automated scoring, automated recommendations, and open-source availability.

DQPA’s architecture is deliberately engineered around a strict separation between deterministic computation and generative artificial intelligence. The platform is a multilayer web application: a React single-page client, a Node.js and Express server exposing a REST API and hosting the deterministic assessment engine, a MongoDB document store accessed through Mongoose, and an external large language model service invoked only after results have been computed. Authentication is token-based with role-based access control, and both tiers deploy as independent containers. Crucially, the normative model itself—processes, questions, weightings, and improvement tasks—is maintained as configurable data through an administration module restricted to expert users, meaning the platform can adapt to revisions of the specifications or even to equivalent frameworks without touching the source code. This data-driven design is what makes the tool reusable and future-proof in a way that hard-coded assessment instruments cannot be.

The heart of the platform is its assessment engine, which algorithmically reproduces the measurement framework of ISO/IEC 33020. Question answers are first converted into weighted percentage scores at two granularities: one per process, from process-specific capability-level-1 questions, and one per capability level from 2 to 5, from cross-cutting questions shared across the scope. The four-category achievement scale maps onto these scores: Not implemented for 0 to 15 percent, Partially implemented for above 15 to 50 percent, Largely implemented for above 50 to 85 percent, and Fully implemented for above 85 to 100 percent. The staged aggregation rule then applies: a process reaches a given capability level when its process-specific score and all lower cross-cutting scores are Fully achieved and the score at that level is at least Largely achieved. The organizational maturity level, on a six-level scale from 0 to 5, is derived from the consolidated capability of the assessed processes in a staged manner. The entire computation is deterministic, traceable, and executed without any intervention from humans or the AI service—a design choice the authors argue is a property rather than a limitation, because transparency and auditability are explicit requirements of the ISO/IEC 33000 method itself.

Only after the numbers are settled does artificial intelligence enter the picture—and the boundaries are strict. The recommendation service neither trains nor fine-tunes any model. Instead, a pre-trained language model, Gemini 2.0 Flash Lite, selected after a structured comparison against alternatives including GPT-4o and GPT-4.1 nano for its large context window and high throughput, is conditioned at inference time by a purpose-built structured prompt. The model’s grounding is entirely deterministic: its input consists solely of the computed as-is state—levels, ratings, and gaps—and a catalogue of predefined improvement tasks curated by domain experts from an anonymized corpus of real projects, assessment reports, standards, and technical documentation. The prompt explicitly forbids the re-computation of levels and imposes prioritization criteria including impact on maturity, criticality of the gap, dependencies, feasibility, urgency, and normative alignment. If the model call fails, the service degrades gracefully to a template-based ordering of the curated catalogue, so a usable plan is always produced without the generative component.

The platform was validated in striking fashion against a real organization: a Spanish public river basin management body responsible for hydrological data acquisition and exploitation. Questionnaire responses were recorded in parallel through DQPA and through the manual procedure of an external expert assessment team, with each business process owner completing the instrument in roughly two and a half hours with support from assessors. The engine computed a largely achieved process-specific score, a partially achieved level-2 dimension, and an unattained level-3 dimension, yielding maturity level 1—precisely the level determined independently by the human experts. Verification went further: the engine’s logic was exhaustively checked against a reference spreadsheet used by the consulting team in professional practice, across all 1,024 possible rating combinations, with full agreement in every case, including boundary conditions between capability levels.

The quality of the AI-generated recommendations was then scrutinized by four expert evaluators—three of them external to the author team and blind to the study—who rated thirty recommendations on a five-point rubric covering consistency with the computed state, alignment with the specifications, actionability, and clarity for non-technical profiles. The overall mean was 4.35 out of 5, with no two evaluators differing by more than one point on any of the 480 ratings, and the restricted external-only mean of 4.23 confirmed the result does not depend on the internal rater. The platform was also applied to three further organizations—a local public administration, a port authority operating critical infrastructure, and a large private technology corporation—producing consistent operation across markedly different sectors, with resulting maturity levels ranging from 0 to 1. Perceived usability, measured with the System Usability Scale across four participants, averaged 80.6, comfortably above the scale’s commonly cited average of about 68. Performance testing showed the deterministic engine computing results in a median of 105 milliseconds and sustaining 150 concurrent users, while end-to-end report generation took a median of three seconds, dominated by the external model call.

The implications reach well beyond convenience. For public administrations and resource-constrained organizations facing obligations under the European Data Governance Regulation, DQPA substantially lowers the barrier to understanding and improving their data practices without depending on scarce certified assessors. For researchers, the platform’s elimination of inter-assessor variability in the computation phase opens the door to genuinely reproducible, longitudinal, and sector-level empirical study of data maturity—questions such as which processes systematically act as bottlenecks, or how maturity evolves after improvement plans are applied, that have been nearly impossible to address empirically until now. The authors are careful to delimit their claims: the platform does not verify declared evidence, cannot prevent deliberate misstatement, does not replace third-party certification, and its recommendations must be contextualized by each organization since the AI knows nothing of internal budgets or politics. Future work includes ablation studies contrasting catalogue-anchored with unconstrained generation, a self-hosted AI deployment for stricter data-residency control, sector-level benchmarking, and what-if simulation of improvement scenarios under explicit resource constraints. But the core message is already clear: the machinery once reserved for expensive consulting engagements has been reproduced as transparent, inspectable, open-source software that any organization can pick up and run.

Subject of Research: An open-source software platform for ISO/IEC 33000-based self-assessment of data governance, data management, and data quality maturity

Article Title: DQPA: A software platform for ISO/IEC 33000-based self-assessment of data governance, data management, and data quality maturity

Article References: Gualo, F., Ayuso, Y., Caballero, I., & Piattini, M. (2026). DQPA: A software platform for ISO/IEC 33000-based self-assessment of data governance, data management, and data quality maturity. SoftwareX, 35, Article 103012. https://doi.org/10.1016/j.softx.2026.103012

Image Credits: AI Generated

DOI: 10.1016/j.softx.2026.103012

Keywords: data governance, data quality, data management, maturity models, ISO/IEC 33000, ISO 8000, self-assessment, process capability, large language models, open-source software, SoftwareX, UNE 0080