Platforms vs Purpose-built Solution

Executive Summary:

Industry 4.0 is the watchword. Nearly every global manufacturer is investing in disruptive technology with advanced analytics and AI. As per McKinsey, Manufacturing and plant monitoring, use cases cover a significant 1/3 of this opportunity.

Companies are increasingly weighing the pros and cons of building their apps on AI platforms and using purpose-built software for plant monitoring. While both have their place in digital transformation, which approach is right for your organization?

This article summarizes the experiences from working with process industry supermajors and at least a dozen oil & gas, petrochemicals, and utility companies globally on digital transformation use cases.

Specifically, the article compares and contrasts – Value Proposition, Core Functionality, Availability, Costs, ROI, and Resource Requirements – for both these solutions to provide a practical perspective to help organizations make the right choice.

Plant Monitoring: Opportunity & Challenges

Plant monitoring—specifically asset health and process efficiency—is at the core of every manufacturing plant, offering a potential of up to $50 billion in annual savings. Businesses are increasingly looking for advanced solutions to overcome the restrictions of traditional approaches, including limited equipment coverage, minimal handling of process upsets, significant noise alarms, high expertise required, and substantial manual effort.

As a result, over 70% of the industries have already started pilot projects searching for the next generation solution. The two major categories of solutions include:

AI Platforms & Purpose-built AI Solutions

Significant differences exist between the two solutions, and choosing an approach requires careful consideration of their pros and cons.

What Problem are they Solving?

AI platforms help software, data science, and IT engineers combine disparate technologies to build an analytics application.

AI Platforms offer an integrated environment with pre-built data connectors, transformations, machine learning models, UI development, and DevOps tools to address this. Platforms are tools to build your analytics.

On the other hand, Purpose-built AI solutions are developed for plant engineers to increase the availability and efficiency of plant operations.

They offer pre-built connectors, data transformations, models, dashboards, and deployment options designed specifically for manufacturing industry challenges. Purpose-built plant monitoring software is a solution for your plant’s operational issues.

Availability of Market Solutions

There are many more AI platforms in the market than Purpose-built AI solutions. Every major industrial company and several new software companies (you know who) offer such platforms. The reason is simple—in theory, a platform can solve any use case and serve finance, legal, manufacturing, and others. So, there is more potential to make money, so more solutions are available.

On the other hand, purpose-built AI solutions are only a handful. Purpose-built AI solutions are limited because they require bridging the gap between analytics, IT, domain expertise, and operational knowledge, which fewer companies can bring. While several traditional solutions were built before 2010, we do not consider them as they do not offer dedicated AI functionality and built-in expertise to address the limitations we mentioned earlier.

Core Functionality

In-house Resources

Organizations need an entire product team to build the plant monitoring solution on AI platforms—software, data science, domain experts, product managers, dev ops, and IT. Customers often underestimate the resource requirements, significant differences between IT and OT technology, and challenges in translating plant engineers’ needs into an intuitive plant monitoring solution.

Conversely, Purpose-built AI solutions do not require additional resources other than the plant engineers. Moreover, subscription-based offerings cover all maintenance and support-related activities.

Cost

Building a plant monitoring solution on an AI platform can be a costly endeavour, costing tens of millions on top of the platform costs. The high expenses are due to the complexity of building a solution that requires a full product development team. Moreover, hiring all the resources needed is a significant long-term commitment, even for the largest players.

Further, the sustenance of such solutions is perpetually making the costs significantly more than purpose-built AI solutions.

Purpose-built AI solutions are typically less expensive to start. While many traditional players offer perpetual with managed services, AI solutions generally are subscription-based and cover maintenance and support. Further, subscription-based pricing reduces the cost of proving the value.

Return on Investment

Final Thoughts

AI Platforms and Purpose-built AI solutions for plant monitoring have merits and limitations.

An AI? platform is a do-it-yourself job, whereas purpose-built solutions offer a ready-to-use solution. Unlike more straightforward use cases like sales or demand forecasting, plant monitoring applications are highly complex due to the integration of IT, OT, operational workflows, domain knowledge, changing process conditions, and data science.

As an analogy, developing your plant monitoring application on an AI platform is like building a car from scratch. It is complex and requires significant effort, budget, skill set, and a risk appetite. Pilot projects often do not represent the real complexities involved—like considering a brake change as a representative pilot for making a car. Such a mismatch in the effort explains the pilot purgatory that many companies are experiencing.

Despite the challenges, AI platforms promise to build something unique and a custom IP owned by the organisation. Organizations prepared to justify such investment with a massive appetite for failed AI pilots' risk, hoping for a possible competitive edge, should consider this approach.

For the rest, best-in-breed Purpose-built AI plant monitoring solutions offer the quickest and highest?ROI. However, one needs to carefully evaluate the true AI capabilities as these solutions are pre-built. Many platform-based apps are too generic, data-centric, and lack domain expertise to deliver the benefits of AI. For more information on differentiating true AI solutions and others, contact us for a copy of the white paper on “Difference between Traditional & AI-based plant monitoring?”

Further, organizations may choose to take a hybrid approach to get the best of both worlds. They could use a purpose-built AI solution for plant monitoring, which covers a large part of the digitization opportunity, and build custom applications on AI platforms for less complicated use cases in sales, marketing, HR, finance, and others.

It is not one or the other but choosing the best solution for the right use case.