How manufacturing can unlock the real power of AI

Step 1: Establish an ‘AI value realisation office’ and evolve into a control tower

As manufacturing organisations start with the “act now” mindset, they can set up an AI value realisation office with a focused scope and C-suite accountability to test and learn. That then can “evolve” into a “control tower,” i.e., a C-suite business unit that is tasked with steering cross-organisational initiatives and the AI strategy.

What this means really is manufacturers need to form a unit that streamlines experimentation and resources around AI and ties it to business outcomes. The value realisation office shapes governance and coordinates knowledge sharing, but its primary goal would be to realise benefits, optimise resources, conduct projects, and risk management exercises. The value realisation office can start with a smaller scope but can then extract maximum value from AI initiatives across the enterprise. The control tower would have the potential to manage re-skilling needs, data infrastructure upgrades and ecosystem strategies.

Step 2: Explore future scenarios to align the approach to AI

To devise targeted initiatives in line with the overall AI vision, manufacturers could start with a future-back approach. This would involve assessing the potential influence of AI on the business and the industry. Future scenarios should account for regulatory, macroeconomic, supply chain, and resource limitations, and should connect AI actions with business value.

Manufacturers could optimise resource allocation and prioritise initiatives more effectively by determining scenarios for possible AI impacts and advantages.

Step 3: Develop a workforce reskilling plan

The growing presence of AI will necessitate reskilling of workforce - both to achieve proficiency in its usage and to amplify competencies that will gain significance in the GenAI era.

Manufacturers can “act now” by driving skills evaluation initiatives to identify re-skilling needs. They could begin with an analysis of tasks likely to be handled by AI and the necessary competencies for workers in areas where GenAI's impact is less studied. They can then “evolve” by formulating and executing a re-skilling strategy.

It is imperative for manufacturers to cultivate a culture of perpetual learning to accommodate ever-evolving skill requirements. They could maintain the engagement of talented workers and lure new talent possessing sought-after skills by creating innovative hubs.

Step 4: Create a data architecture assessment and upgrade roadmap

Performing an evaluation of their data architecture and then implementing a data strategy with upgrades based on ROI, impact, and feasibility could help manufacturers navigate the journey from “act now” to “evolve later.”

To implement GenAI / advanced AI throughout the workforce, large language models (LLMs) need to be trained on standard operating procedures and best practices, thereby creating a “knowledge graph” for the organisation. However, a significant amount of this information typically exists solely in the minds of employees and is often neither formally codified nor digitised.

The data architecture needs to be appraised to determine process design, data quality and security, and dependencies. Thereafter, suitable benchmarks could be put in place to create a performance baseline and to bolster future AI business cases. And once the infrastructure is mapped, manufacturers would need a strategy to gather, save, and administer the data necessary for AI applications. This could help with demand forecasting, financial forecasting, and manufacturing risk to improve business efficiency.

Step 5: Develop AI ecosystem partnerships

Manufacturers are used to handling intricate supply chain partner ecosystems. It’s, therefore, important to set performance standards, manage partnership costs, and vet partners.

To “act now,” manufacturers need to chart AI ecosystems and associated capabilities and commence pilot programmes. And then as they “evolve,” they can employ key evaluation metrics to foster a compact yet robust network of AI partners.

To gain experience in AI projects, a good step forward is to form initial partnerships with multiple entities and create opportunities for small pilots. As the AI partner ecosystem matures, it’s crucial to define key metrics for assessing ecosystem relationships. This would help in selecting and nurturing relationships with high-priority partners while decisively eliminating redundant partners.