Karen Kim, CEO Human Managed

While the hype around Gen-AI will shift to the buzz around agentic-AI in 2025, traditional sectors such as manufacturing will still be coming to terms on deploying AI strategies to become participants of Industry 4.0 or the Fourth Industrial Revolution.

This is the next phase where disruptive trends including data, connectivity, analytics, robotics and human-machine interaction is said to transform manufacturing significantly.

According to a Kearney study[1], while over 70% of ASEAN manufacturers polled claimed to be “non-leaders/followers” of AI adoption, 47% believe investment in AI is “very critical” for their business. Gartner’s[2] Prism use case of GenAI in Manufacturing encourages  leaders to evaluate AI use cases on a strategy that addresses both value generation and feasibility .

• Value parameters include increased revenue, increased efficiency, managed risk and non-financial gain.
• Feasibility parameters include technical feasibility, internal readiness, external readiness.

So how can leaders operationalise AI, keeping value and feasibility in mind? At Human Managed, we believe it begins with an understanding of data. A framework to understand the highest impact for manufacturing businesses in the age of AI, would be to understand “3 Ops”-DataOps (Feeding AI), MLOps (Tuning AI), IntelOps (Applying AI).

Data Ops (Feeding AI)

In the manufacturing sector, data can be provided through assets in an interconnected ecosystem. From a range of sensors, machine data on temperature, vibrations, pressure, power consumption, operating hours etc are collected. Industrial machines, robotic arms and assembly lines can provide process data includes information on production rates, cycle times, error rates, material usage. Image and video data from cameras capture data for product dimensions and defect detection. ERP systems provide information on inventory levels, order information, transportation data.

Given the vast array of data, manufacturing enterprises have to first solve for the issues around understanding the data itself. These can be clubbed into 5 buckets –

1. The visibility problem – to have a current view of assets, their posture and behaviours to be able to track business changes.
2. The verification problem – to ensure that data outputs generated by tools and solutions are verifiable and trusted.
3. The foresight problem – to predict accurately based on large volumes of reliable data, effective models and quick feedback loops.
4. The prioritization problem – to prioritize quickly, which data, events and alerts are essential and need attention.
5. The orchestration problem – to bring together multiple tools and technologies into cohesive operational pipelines to detect, react and respond effectively given the volumes and velocity of increasing data.

 ML Ops (Tuning AI)

After establishing the ground reality from a data perspective, manufacturing enterprises need to zoom out to see opportunities in business models from an AI modelling perspective. Manufacturing data can train machine learning models for a vast array of use cases including predictive maintenance, quality control, supply chain optimization, process optimization, robotics and automation.

While there have been significant advances in foundation models and large language models,  with many plug-and-play AI tools, operating AI at scale across multiple use cases and business processes, requires custom-trained models. However, this is also where most MLOps and LLMOps challenges occur, such as:

1. Choice paralysis : with AI developments evolving so quickly, deciding what direction to take and what models to use is challenging.
2. Lack of technical expertise : even after identifying use cases and models, deep technical knowledge and capacity for experimentation is needed.
3. Siloed data flow : to achieve holistically contextualised intel from data, data sources and management which are typically isolated in enterprises, need to be integrated.
4. No context models: For machine learning to be operational, AI models must be trained, tuned, and improved with data, logic, and patterns unique to your business.
5. Feedback loop: AI models only improve with proper feedback, so decisions must be made about who will give the feedback and the associated mechanism. There is also a responsibility to check for and reduce bias and prejudice.

IntelOps: Applying AI

Even when you make headway with the DataOps and MLOps problems, one big piece of the puzzle is left in operationalized AI: presenting and serving the AI outputs to the correct recipient (humans or machines) at the right time.

For manufacturing enterprises, these could include real-time monitoring and alerts to track machinery, production and supply chain in real time, predictive analytics for planning and resource allocation, integrating AI-driven analytics to support decision in quality control, inventory management and workforce allocation.

While the outputs of DataOps (insights & intel) and MLOps (labels) are important, what is also essential—but often deprioritized—is how they get generated, delivered, and tuned to be usable across the business.

At Human Managed, we call this continuous process of data-to-intel, intel-to-labels, and labels-to-serving pipelines: IntelOps. To get the most out of AI in your day-to-day business processes, it’s crucial to build in the specifics of the distribution of labor between humans and machines.

• What Human or AI processes analyse what type of data and use cases? (Logic, model)
• What Human or AI processes generate outputs (insights, recommendations)?
• What Human or AI processes do you use to execute actions? (Functions, tasks)
• Finally, how and where do you present the AI outputs? (API, report, notification)

There is no one mode of AI; it will be different based on the type of implementation, function and the level of automation required.

Conclusion: Balancing value and feasibility of AI in manufacturing depends on data, context, and processes

The future of AI in manufacturing will belong to companies that prioritise building a robust, data-driven foundation, making data, models and their outputs operationalized and always production-ready- not just once, but every day

By breaking away from reliance on isolated, human-held knowledge and fostering collaboration with a distributed ecosystem of partners and suppliers, these forward-thinking manufacturers will not only harness AI to enhance efficiency but also gain a strategic edge in innovation, making Industry 4.0 a reality.

We believe that the companies that continuously build the context of their business as distributed and scalable data (instead of tribal knowledge in individuals’ minds), and work with a distributed ecosystem of partners and suppliers, will be the ones that grow with AI, balancing dual parameters of value and feasibility.

[1] Kearney Study

[2] Gartner Prism GenAI Manufacturing

CEO, Human Managed
Karen Kim is the CEO of Human Managed, the ASEAN cloud-na ve data & AI platform that empowers
businesses to make smarter decisions and faster ac ons for cyber, digital and risk outcomes.

Karen’s responsibility at Human Managed is to align its purpose to strategy, and strategy to “Distributed
Ops” for data-driven solu ons, including DataOps, MLOps, and IntelOps. She enjoys applying her
learnings, love for design-thinking and service-first mindset to various domains like branding, service
design, and business development. She is proud to lead a “different kind of” company that dares to solve
complex problems of today’s data-flooded world.

Karen speaks and writes frequently on topics ranging from digital transformation, cybersecurity, and
AI to organisational culture and leadership. Her insights have been covered extensively in the media
across ASEAN, including e27, CDO Trends, The Philippine Star, Business Today Malaysia, Tech
Collective Asia, FedEx Insights, HRO, How To Live Podcast, Yahoo Finance, Uptech Media.

About Human Managed

Human Managed is a cloud-na ve data and AI platorm that empowers businesses to make smarter
decisions and faster actions for be er cyber, digital and risk outcomes. The company helps
enterprises of all sizes organise and get visibility of their data, serves them personalised intelligence
to make be er operational decisions, helping them overcome problems of information overload,
siloed technologies, and legacy architecture.

The company’s proprietary Intelligence Decision Ac on [I.DE.A] platform, built on 50+ out of the box
integra ons with leading technology providers, models intelligence specific to each customer. The
personalized IDEAs are delivered as reports, notifications, and dispatch for fast and actionable
decision-making. Taking data from any source, the platform offers customised, pay-as-you-go
modularity to clients, via 14 separate engineering functions and 92 microservices.

Human Managed has been awarded the ISO/IEC 27001:2022 certification for its services in security
opera ons, data science, and incident response coordination. It is also SOC 2 Type I compliant.
Founded in 2018, Human Managed is headquartered in Singapore and operates across the
Philippines, India, and Hong Kong. For more information, please visit:
https://www.humanmanaged.com or follow them on LinkedIn.