CLOSING THE LOOP IN PHARMACEUTICALS INDUSTRY FOCUS Medical & pharmaceutical
from all of these sources and deliver timely, actionable business and operational insights. Chanakya is an AI-powered expert advisor
Balajikasiram Sundararajan, Chief Digital Officer (CDO) of ACG World, says smart manufacturing delivers when digital systems, automation infrastructure, and human expertise come together to close the loop
O
n the manufacturing shop floor, closing the loop delivers tangible results. For example, Overall Equipment Effectiveness (OEE)
improves when unplanned downtime is reduced through targeted maintenance interventions, and defect Parts Per Million (PPM) drops when process corrections are specific and timely. Historically, these actions were guided by operator expertise, Standard Operating Procedures (SOPs), and rule-based prescriptions. With advancements in decision sciences, AI, and connected automation systems, it is now possible to generate actionable insights that translate directly into shop-floor interventions. The following two use cases from our manufacturing plants demonstrate this. Golden Batch – first-time-right production in pharmaceutical capsules manufacturing We manufacture pharmaceutical capsules
across thousands of colour variants. Production speeds run into several thousand capsules per minute per line, and dimensional tolerances are measured in microns. Our capsules are produced on machines of varying vintage and operational condition, from brand new to over twenty years old. Each machine has more than 60 critical settings that directly impact capsule quality and yield. Traditionally, machine setup was guided by operator expertise and SOPs. However, the sheer variety in our system - product variants, machine conditions, and parameter combinations meant that manual methods frequently led to multiple setup runs resulting in lower first-pass yield, higher defects, and higher capsule losses. To solve this, we developed Golden Batch, a system that prescribes optimal values for all 60+
32 May 2026 | Automation
critical machine settings using an ensemble of machine learning models trained on three years of production, operational, quality, and machine condition data. A continuous learning pipeline refines these models based on production outcomes and operator feedback. Once the operator confirms the recommended settings, they are transferred directly to the machine Programmable Logic Controller (PLC), closing the loop from data- driven insight to machine-level execution. Golden Batch is built using Industrial IoT, machine learning, edge computing, and cloud computing technologies. Today, Golden Batch is deployed across our capsule manufacturing plants, and has delivered a significant improvement in first- pass yield, reduced capsule losses, shortened setup time, and lowered defect PPM. Chanakya - enabling shop-floor operators with actionable recommendations in pharmaceutical packaging materials manufacturing
Our packaging materials business has 5000+ SKUs, micron-level substrate tolerances, very short order visibility, more than 30% rush orders, and delivery lead times in days. Our operational decisions need to be fast, well-informed, and grounded in domain expertise. Two challenges made this difficult. First, many of our long-serving experts who carried this knowledge were approaching retirement. Second, the information needed to make high-impact decisions was scattered across multiple disparate systems such as order management, planning, procurement, production, quality, logistics, maintenance, and finance. No single view existed. We needed a solution that could draw
that orchestrates 15+ generative AI assistants aligned to specific functions such as production, quality, and maintenance. For every query, it provides a unified, cross-functional response comprising a direct answer, deeper insights derived from the underlying intent, visual substantiation through charts and graphs, and recommended next steps. On the shop floor, this translates into practical recommendations for supervisors and operators – for example, identifying likely causes of recurring stoppages, highlighting material or planning constraints, and recommending maintenance or process actions. Chanakya is built using generative AI, Model Context Protocol, and cloud computing. Today, Chanakya is live across our packaging operations. It has delivered measurable improvements: higher machine uptime, greater process consistency, reduced mean time to repair, and faster customer response times. These two solutions operate in different industries and close the loop in different ways. Golden Batch prescribes machine settings and transfers them directly to the PLC - the system acts. Chanakya synthesises cross-functional insights and delivers recommendations to operators and supervisors - the human acts. Apart from the complexity of the solutions, multiple model iterations, and other technical difficulties, the key challenge was change management. How do you make an operator or supervisor let go of expert judgement and experience to accept recommendations from a system? Two key interventions helped us make change happen: continuous involvement of shop-floor personnel from solution conceptualisation stage, and gamification for solution adoption comprising points, shop-floor leaderboards, rewards and recognitions. Golden Batch and Chanakya are two examples from ACG’s smart manufacturing solutions portfolio. Two “Global Lighthouse” recognitions from the World Economic Forum are a testament to the company’s journey. Smart manufacturing delivers when digital systems, automation infrastructure, and human expertise come together to close the loop - converting operational data into impactful decisions, and decisions into measurable shop- floor action.
ACG World
www.acg-world.com
automationmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40
