A complex industrial mandate - scenario

Politecnica, a key manufacturing arm of the Arroweld Group, operates in a high-precision sector: the production of custom plastic components and industrial machinery parts. Unlike other group entities that focus on resale, Politecnica creates bespoke products. This operational reality involves high variability. Every order begins with a technical drawing, requiring a specific evaluation of machine time, material costs, and margin. The company had accumulated fifteen years of detailed production data—commissions, drawings, and machine logs—stored within their AS400 archives. However, this massive dataset remained dormant, leaving the quoting process dependent on human experience rather than empirical fact.

The subjectivity bottleneck - challenge

The challenge was not just efficiency, but resilience. The quoting department, staffed by approximately 10 full-time employees, relied heavily on the subjective expertise of senior 'preventivisti' (estimators). The existing process contained specific systemic vulnerabilities:

The existing process contained specific systemic vulnerabilities:

• Reliance on tribal knowledge: The deep understanding of 'how long this part takes' resided in the minds of staff nearing retirement, creating a significant business continuity risk.
• Inaccessible heritage: 15 years of 'smart machine' data (Industry 4.0) existed in the archive but was technically retrievable only through manual, time-consuming searches.
• Variable outputs: Two different operators could assess the same drawing and produce different cost structures based on their personal experience levels.

The objective - expected outcome

The goal was to transition from a subjective craft to a standardized science by introducing a tool that could:

• Automate similarity detection: Instantly identify if a requested part (or a similar one) had been manufactured before.
• Standardize pricing logic: Ensure the price is a result of 'Machine Time x Margin,' derived from actual historical performance rather than estimation.
• Secure knowledge: Digitize the expertise of senior staff so the company's intelligence remains intact regardless of personnel changes.

Precision Quote: Knowledge activation - the solution

The project was executed through a co-creation model involving Politecnica's leadership and digital governance teams. The solution, 'Precision Quote,' is a GenAI-powered platform that bridges the gap between unstructured technical drawings and the structured data of the AS400 ERP. The system is structured around three core capabilities:

Three Core Capabilities

The system is structured around three core capabilities:

• Intelligent Drawing Analysis: The system ingests the customer's PDF or image file, analyzing the geometry and technical specifications.
• Semantic Similarity Search: It queries the 15-year archive not just by name, but by characteristics, retrieving the 'closest match' from thousands of past projects.
• Evidence-Based Costing: Instead of guessing, the operator is presented with the actual machine time and costs from previous similar jobs, allowing for a quote based on proven reality.

Efficiency, control, and strategic transformation - the outcome

The project has resulted in strategic outcomes with significant organizational impact: From subjective art to objective standard: The quoting process is no longer a test of memory. By presenting the operator with historical precedents, the system creates a baseline of truth. This standardization allows junior staff to quote with the accuracy of senior experts, flattening the learning curve and removing the risk of error. Asset reactivation: The AS400 archive has transformed from a passive storage unit into an active strategic engine. Data from 15 years of production is now the primary driver of current revenue, validating the ROI of previous Industry 4.0 investments. Strategic 'Servitization': Beyond efficiency, this shift enables a new business model. Politecnica is moving from simply selling parts to selling production capability as a service. The speed and reliability of the new system allow for a more responsive, service-oriented relationship with customers.

Lessons learned: a model for pragmatic AI

The project established an operational model for introducing GenAI in traditional sectors:

Key lessons learned:

• AI as continuity, not just automation: The most critical value of AI here was not replacing people, but preserving their knowledge. By embedding human expertise into the system, the company immunized itself against the loss of key talent.
• Innovation respects legacy: The project succeeded because it did not attempt to replace the core ERP (AS400). Instead, it built a modern intelligence layer on top of it, extracting value from legacy systems without disrupting the operational backbone.
• Focus on the 'Job to be Done': The solution avoided the trap of 'technology for technology's sake'. It focused entirely on a single, high-value business problem: the quote. This specific focus ensured high adoption and immediate measurable impact.

Key insights for your organization

The experience with Politecnica provides a replicable roadmap:

• Don't ignore your archives: Your most valuable AI training data is likely already sitting in your legacy storage.
• Standardize to scale: You cannot scale a business if its core processes rely on individual memory.
• Bridge the generation gap: Use technology to transfer the 'tribal knowledge' of retiring experts to the systems used by the next generation.