Multi-Site AI Agile Coaching Across a Penang Headquartered Semiconductor MNC

The situation

An enterprise of this profile is a semiconductor multinational with its Asia Pacific operations headquartered in Bayan Lepas or Batu Kawan, with additional sites in Kulim and Johor. The footprint covers assembly, test, advanced packaging, design verification and a growing share of system-level engineering. Production reliability and yield are the working language of the floor. Engineering managers are technically credible. The corporate group is one of the larger players in the global E and E ecosystem and is under pressure from the parent organisation to embed AI into the operating model without losing the production discipline that the Malaysian sites are known for.

The participant population on a typical engagement is between one hundred and three hundred across the four sites. Engineering managers, senior process engineers, manufacturing engineers, supply chain leads, the embedded quality engineering function and the site IT and digital teams. The Penang Skills Development Centre regional network is part of the talent context. The Pembangunan Sumber Manusia Berhad framework is part of the funding context, since the company is HRD Corp registered and uses SBL Khas for claimable training spend.

The business pressure is the same pressure the global E and E sector is under. AI is moving from a corporate strategy slide to a floor expectation. The local engineering managers are asked to land AI augmented yield improvements, AI augmented defect classification, and AI augmented predictive maintenance, while keeping the operating system intact. Engagements of this archetype begin when a regional engineering vice president decides the coaching budget cannot remain ad hoc.

What we observed

1. The engineering managers are stronger than the AI vendors assume

External AI vendors often treat the floor as a deployment surface. The floor is in fact a sophisticated engineering organisation that already runs structured improvement loops. The coaching has to respect that. Coaching language that talks down to engineers is rejected, correctly.

2. AI use is uneven across sites

One site has a pocket of advanced AI work. Another has none. A third has experiments that were abandoned. Without a multi-site cadence, the company cannot learn from its own experience.

3. The improvement loop is the coachable artefact

Engineering improvement loops are the most coachable artefact on the floor. Rewriting them to include AI augmented hypotheses, AI augmented data work and AI augmented retrospectives is the highest leverage move available.

4. The site IT and engineering relationship needs work

Site IT and engineering are too often parallel functions. AI work cannot land while the two functions are negotiating ownership. The coaching has to bring them into a shared cadence.

The engagement approach

Engagements of this archetype run as a multi-site programme with founder-led discovery and senior-trainer coaching at each site.

Multi-site discovery, month zero

The founder, Prashant Shinde, an HRD Corp Train-the-Trainer certified founder, visits each site. Output is a one page site profile per location, a shared multi-site target operating model, and a sponsor map naming the regional VP, the site directors and the site engineering managers who will hold the coaching cadence.

Site intensive, months one to three

Each site enters an eight week intensive. Engineering managers and senior process engineers take a coaching track based on ICP-ACC and the AI Agile for Penang Semiconductor module. Senior engineers and team leads take an AI productivity and workflow redesign track based on AI Productivity Foundations. Both tracks bring real improvement loops as the work product.

Multi-site community of practice, months four to six

The site community of practice is then connected across the four sites. The founder runs a monthly senior review where each site presents one rewritten improvement loop. Senior coaches travel between sites to keep the standard.

Embed and certification, months seven to nine

The senior engineering managers who will continue the cadence complete ICP-CAT. The community of practice is now owned by the regional VP and run by a site sponsor on rotation. Agile Visa moves to a quarterly senior reviewer role. The Pembangunan Sumber Manusia Berhad funding documentation is closed for the year.

What changed

1. Improvement loop quality across the sites

The rewritten improvement loops are demonstrably stronger. The site that had no AI use is now contributing pattern entries to the multi-site library. The site that had abandoned experiments has restarted them with better discipline.

2. Engineering manager confidence

The engineering managers can hold a conversation with the regional VP about AI without ceding ground to vendors. This is qualitative and decisive.

3. Time from AI hypothesis to floor trial

The time between an engineering manager identifying a candidate AI augmented improvement and a controlled floor trial reduces. The reduction is sustained, not a one-off.

4. Multi-site learning velocity

The company starts to learn faster from its own four sites than from external benchmarking. This is the strategic prize.

Lessons that transfer to other Penang and Klang Valley E and E MNCs

• Respect the engineering operating system. Coaching language that ignores the existing discipline will be rejected by the floor.
• Build the multi-site cadence early. Without it, every site re-learns the same lesson.
• Bring site IT and engineering into one cadence. AI cannot land while ownership is unresolved.
• Coach the improvement loop, not the tool. The loop is the durable artefact. The tool will change.