Artificial intelligence has become the latest buzzword dominating enterprise software. Dreamforce 2016 notwithstanding, Google searches for AI have steadily increased over the last few years and really spiked in just the past few months.
According to this Forbes article, IDC expects that close to half of all enterprises will elect to create some form of an advanced digital transformation. GE Digital, which is acquiring ServiceMax, recently announced the acquisition of AI company Wise.io.
What’s the appeal? Well, for many CRM or ERP platforms that do a good job already of digesting massive amounts of customer and company data, AI promises to essentially do an even better job and take manual analysis out of the equation. Presto: better decision-making.
Over the past few weeks we’ve seen a number of companies demonstrating the possibilities and vendors waxing poetic on the promise of AI and 80 percent of tech leaders, according to this study, posit that the net effect of AI will be that it “creates jobs and improve(s) worker performance and efficiency.”
The field service industry, as many may know, has been undergoing massive transformation itself. Internet of Things initiatives make it possible for companies to remotely monitor and proactively maintain industrial equipment, saving millions in unplanned downtime and service costs; the mobile cloud has equipped a legion of field service engineers with a treasure trove of customer information and automated processes wherever they go. Augmented reality is also making a good case for field service applications.
And now AI is starting to make its way into the field-service-transformation conversation – and for good reason; there are a number of ways field service could benefit from AI.
But the reality is it’s not a simple plug-and-play operation.
AI’s success is in general dependent on the data being fed into it. Just look at the evolution of data processing techniques and technologies in the enterprise world.
In “AI’ing” field service, unfortunately, this industry sector brings the additional challenge of not having a lot of good, existing data for the normal application of analytics and AI. Furthermore, even when there is data available, it comes is all shapes.
- Business Processes & Data: While in general, data has crucial insights to offer, the magnitude, depth, and heterogeneity of it gets multiplied in the context of field service. No two field-service organizations execute their field service operations exactly the same way. Each process has its own unique way of handling data. For example, efficiently planning and getting visibility into inventory is so different from efficiently managing and dealing with a “Bill of Materials” as maintained. And then, data flows into the system in all forms, from fully structured to semi-structured to fully unstructured. It is important to ensure that the dots are connected end to end, and data is understood in a context specific way and insights are derived that can cater to both for process optimization as well as business optimization purposes.
- Actors: Individuals with different roles come in and go out throughout the flow. These users can be technicians in the field with mobile devices, back office users, dispatchers, service owners, and so on. With IoT, now machines are connected to the enterprise systems and drive data flow as well as business processes. Each role has unique job complexity, understanding of the system, and the expectations from it. There is a need to make the engagement across all stakeholders efficient, productive, and optimized based on continuous feedback mechanisms.
- Decisions: Users make decisions as data flows through the process. Decisions are either automated via rules, either semi-automated or explicit. The accuracy of decision-making in many ways impacts key field service metrics that indicate the evolution (or the lack of) of a field service organization. An intelligent IoT system’s ability to always eliminate noise and trigger one or more business process only when it is warranted; or a technician carrying all the parts he/she needs when visiting a customer site to fix an issue with IB; or a field technician looking for help. There are decisions that are being made and ideally need to be automated/system aided. Most importantly, the decisions should be made by an intelligent system, even if it is artificial.
And while enterprise data is truly king, it’s simply difficult to make computers understand data the way people do. The hardware to support it currently costs too much and doesn’t perform well enough; programming technologies lack sophistication.
As you can see, for some domains and specifically field service, AI cannot be a one-trick pony. To create a deserving impact, AI has to be seen and applied holistically to field service. We need to have a long-term vision to make field service smarter than it is today. Some potential areas that could contribute to a holistic AI strategy for field service include:
- Automation: If a manual intervention is not required, then eliminate it. A rule-based automation approach is a good start but cannot scale after a point. We should also stop automating a single task and instead look at automating an end-to-end job. AI is better suited for that.
- Process Evolution: Businesses often choose a practical starting point to get going when they deploy a new business process. It is important to monitor and measure how the new process is working (or not working). This feedback is essential in order to evolve the process and reach a close to optimal working condition.
- Autonomous Field Work Force (Not to be confused with autonomous systems): Extending on the concept of decision making, allowing the field workforce to manage its work can greatly increase the efficiency and productivity of the workforce and positively reduce the dependency on explicit management. The key is to make sure the system engages with the workforce throughout to help them manage themselves.
- Outcome-First Machine Learning: A big advantage of machine learning over AI is the actual historical data itself can be optional, as long as we have ‘quality’ training data that can lead to an expected outcome. This is can be a great starting point for Field Service organizations who want to leverage the new technology innovations like predictive analytics but have not maintained historical data in a consumable fashion (worst case scenario is pen and paper).
- AI On The Go: One of the key roles within any Field Service organization is a Field Service Technician. It means that the impact of AI adoption is not even close to complete if the benefits of AI are not delivered to the mobile devices. Furthermore, AI should continue to assist independent of whether the device is connected to the Internet or not.
- Natural User Experience: A moderately complex challenge that a field service technician faces today is how the mobile application helps him/her work. And it is far from ideal. More focus should be on the work itself and less on how you capture it. Things like natural language processing, speech recognition and wearables can drive a natural, intuitive, and truly hands-free user experience.
- In the “Line of Sight” AI: AI is not an add-on. It cannot be a number in some column that you have to interpret OR a tool that you to run elsewhere away from the primary system. The key to delivering AI that is to build it into user’s “usual” workflow.
- Assistance: Key to implementing effective automation and an autonomous workforce is the level and precision with which the system guides the users. Context specific help and smart collaboration techniques can be of great help.
- Service Intelligence & Insights: Obviously, it’s the whole point. Driving smart decisions in all the key areas of field service: Parts & Inventory, Warranties & Contracts, IB Management, and Work Orders, just for starters.
A Word of Caution
It’s easy to get carried away. AI and machine learning are such fascinating approaches to next world solutions, companies jump on them too quickly without understanding the complex nature of the approaches, wider applicability and sometimes potential for short-term gains.
It’s also a mindset change that both software vendors and businesses have to go through to get over the “conventional” problem solving approaches. Unless you have the power to bend the space, there is no shortcutting the time it takes to get there.
“AI has to be seen and applied holistically to field service” – absolutely true. I am not an expert but the “field” in field service is where the complexity lies. With all the dynamics involved, it may take time for AI to be embraced by field service, but how and when will be intriguing
Artificial intelligence will always be limited by the lack of being ale to perceive all of the relevant information. It will also be terribly limited by not being able to understand how a system works, which is an incredibly handy asset when working to get a system that has failed working again. Any AI that was complete enough to fully understand a system, and the effect of a failure in any part of that system, and the required service action to restore the system to correct function, would be “huge” in both value and complexity.
But, since those creating the AI seldom are familiar with the systems to be serviced, it is unlikely that such will ever be created.