Manufacturing SMEs: when the smart factory and AI move from theory to reality
a collaboration between kuriosIT and Baseline
KuriosIT and Baseline combine their expertise to offer an informed perspective on digital transformation in the manufacturing sector.
Part 2
In our previous blog, " Manufacturing SMEs: smart factory and AI, a winning duo ", we have established that data reliability is the essential foundation for any artificial intelligence initiative. Indeed, systems need to "talk" to each other to transform raw data into strategic intelligence.
In this blog post, we take the next step. We look at the following three concrete use cases inspired by our realizations:
IoT data architecture: the foundation of manufacturing AI projects
In many plants, AI projects fail not because of the models or algorithms, but because PLC, SCADA, MES, and other system data are not structured properly. Without an IoT foundation, AI cannot transform this data into reliable, actionable decisions.
Challenge: Hard-to-use data
In most plants, floor data comes from a stack of systems (e.g. PLC, HMI, SCADA, MES, quality systems, CMMS) installed at different times. These systems work, but they were never designed to feed advanced analytics or AI models. As a result, we accumulate a lot of data, but very little of it is actually usable.
Scattered data: Information is distributed across several systems without a common data model.
Non-standardized naming: Each supplier or integrator uses its own tagging conventions.
Lack of context: Data is rarely linked to equipment, products, batches, recipes or shifts.
Inadequate logging: Data is either missing, or kept at a granularity unsuited to advanced analysis.
Break OT-TI: It's difficult to link production data to business systems such as ERP, quality or inventory.
Solution: Create a reliable, consistent IoT architecture
To make plant data usable by AI, it is first necessary to create a structured and coherent IoT architecture. The aim is to centralize, standardize and contextualize data right from its source, so that it can feed advanced analyses and predictive models.
Data centralization: Bring together data from PLC, SCADA, MES and other systems in a single platform.
Tag standardization: Harmonize naming conventions and units to make it easier to compare information.
Contextualization: Add information on equipment, product, batch, recipe or shift to enrich analyses.
Customized archiving: Maintain data with the granularity required for different types of analysis and AI models.
OT-TI integration: Gradually link production data to ERP, quality and maintenance systems to obtain a complete, consistent view.
Benefit: Leverage industrial AI with confidence and organized data
Once the OT foundation is in place, AI delivers more value. Models leverage reliable, consistent and contextualized data, facilitating plant-wide decision-making and deployment.
More reliable analysis: Models and reports are based on consistent, reproducible data.
Time-saving: Less time spent cleaning or correcting data to make it usable.
Team confidence: Operators, managers and engineers can rely on the recommendations generated.
Deployment at scale: Foundation models can be applied to several lines or plants.
Return on investment: AI projects deliver value faster and in measurable ways.
Dynamic allocation of carpet production and operators
This use case illustrates how AI solves the puzzle of planning and scheduling carpet production, optimizing the use of equipment and manpower.
Challenge: Bottlenecks and costs of changing setups
Carpet manufacturing involves a complex sequential flow (Extrusion → Tufting → Kiln → Finishing) often spread across several separate production units. This challenge is exacerbated by tangible factors and constraints that cripple manual planning:
Flow complexity and problem size: The production flow is complex, with multiple constraints. With infinite combinations of possible production schedules, the planner struggles to satisfy all constraints and optimize production.
Critical bottleneck (the oven): The oven is the main bottleneck, requiring long set-up times to switch from one type of product to another.
Extremely costly set-up times (tufting): The tufting stage requires tedious bobbin changes, making the minimization of set-up times critical to profitability.
Resource micromanagement: Static allocation of operators between manual finishing stations and automated cells leads to inefficiency and wasted time.
Unsuitable software: Conventional planning and scheduling software (ERP/MES/APS) often attempts to solve all variants of planning and scheduling problems. Contrary to popular belief, this makes them ill-suited to real-life industrial problems.
Static allocation and manual planning are inadequate to manage these complex interdependencies.
Solution: AI at the heart of scheduling
AI tackles this challenge by building a global vision and using a tailored approach integrating several combinatorial optimization techniques (e.g. constraint programming, wide neighborhood search) to find the optimal production sequence:
Global orchestration: The AI intelligently explores combinations that minimize total production time and setups. It favors tufting campaigns (several mats of the same type) to reduce set-up times that don't add value to the process.
Feasibility guarantee: AI guarantees a valid solution that respects the production flow and all operational constraints, while minimizing setup times.
Dynamic allocation and real-time data: The system integrates industrial data in real time to generate a continuous plan that optimizes the use of equipment and operator tasks (e.g. switching between manual cutting and automated finishing).
Targeted approach: The aim is to use AI to solve complex, targeted problems, and then combine the solutions together. For example, an optimized production schedule makes it much easier to assign operators to different machines, and optimizes time management.
Benefit: Reduced hidden costs and increased throughput
This approach makes it possible to become more agile, a decisive advantage:
Tailor-made approach: Unlike generic APSs, this solution is precisely adapted to the real industrial problems and specific constraints of carpet manufacturing. It generates far greater savings.
Maximizing capacity: Significantly increase production throughput by maximizing furnace uptime (critical bottleneck) and reducing non-productive downtime.
Reducing the cost of change: Significantly reduce operating costs by minimizing long, costly setups, translating directly into higher margins.
Effective personnel management: Major facilitation of assignment, scheduling and workforce management through improved production predictability.
Agility in the face of the unexpected: AI generates and adapts the complex schedule in a matter of minutes, making it possible to react to new orders or unforeseen breakdowns.
Assistance with customer submissions
Challenge
For small and medium-sized manufacturers, preparing a customer quotation is a long, complex and painstaking process. Each request imposes precise technical requirements, material specifications, and the analysis of various drawings (e.g. AutoCAD, PDF). This process faces several major obstacles:
Intensive manual search and information fragmentation: Gathering the necessary information (historical submissions, reference manuals) is extremely fragmented. This requires manual examination of dozens, if not hundreds, of documents, consuming precious time.
Dependence on critical human expertise: Correctly interpreting requirements, assessing risks and drawing up accurate specifications all require considerable expertise. This dependence makes the process vulnerable.
Loss of strategic knowledge: With the departure of key employees, expertise and critical data on past projects are often lost. This knowledge drain affects the accuracy of future estimates and the ability to replicate past successes.
These combined challenges slow down the sales cycle significantly and limit the volume of bids the team can handle, thus stunting growth.
Solution: RAG artificial intelligence
AI, and in particular RAG approaches, transforms this process by exploiting the company's documentary heritage. In concrete terms, it aims to:
Universal access to data: The RAG system quickly identifies and structures historical bids, profitable projects and technical documents, whatever their original format (e.g. PDF, JPEG).
Targeted recommendations: The AI then uses these elements to automatically generate a detailed draft bid and make a predictive profitability analysis.
Profit
The main benefit is a drastic acceleration in the generation of submissions and manual document searches. This approach also makes it possible to :
Reduce preparation time and respond to a greater volume of requests.
Improve the accuracy of estimates based on comprehensive data.
Standardize the format and quality of submissions.
Increase mandate rates and maximize targeting of relevant offers.
Build on previous projects and ensure knowledge retention.
Two complementary areas of expertise, one vision
KuriosIT and Baseline are a natural alliance for Quebec manufacturing SMEs looking to make their digital shift a reality. Together, we form a duo that covers the entire value chain: from data collection and structuring, to intelligent data exploitation through concrete AI solutions. Our approach is responsible, iterative and focused on the business objectives and realities of SMEs. We build together, at your pace, without vague promises, to bring you tangible results.
