By Constant Ondo
Field Service Management (FSM) is a use case of PICC Software for which demands keep increasing. Nothing really surprising since service is becoming a strategic issue, or making its comeback as such, due to digitalization and changes in business models.
Companies are considering PICC Software as a solution to improve their troubleshooting and maintenance services because our platform is able, with its artificial intelligence, to combine data coming from IoT sensors and knowledge embedded in experience feedbacks to solve complex problems.
Our aim is to provide augmented intelligence to on-site technicians, so that they can fulfill their mission autonomously, on the first try, whatever the problem is.
Let’s take the example of a technician facing an hydraulic system with a pump that makes a strange noise. With a standard software, he will search within all the procedures related to « pump » the one that seems the most appropriate. With an automatic diagnosis system, he will make a few measurements and/or fill a questionnaire to retrieve the proper procedure. With a software as PICC Software, he will be guided, step by step, until the problem is solved, using a dynamic flowchart.
At each stage, the software will be able to automatically collect condition data from an IoT sensor and, depending on the result, adapt the procedure or not. At each stage, the technician will also be able to insert a comment. Analyzing this comment, the software will possibly detect that the situation does not fit with the ongoing procedure. The technician will be offered alternative solutions, ranked and associated with a risk/benefit analysis. He will not need to call for remote assistance or loose time trying to understand the origin of the problem. He will be able to cope successfully on his own with this unusual situation.
To achieve this result, we are constantly improving the algorithms of PICC Software, using deep learning and the latest developments in semantic analysis. Our latest update is able to reconcile words or phrases which apparently have nothing in common, with a relevance level we wouldn’t have imagined ourselves.
Applied to Field Service, this new cognitive capacity of PICC Software works its wonders !
Deep learning and semantic analysis to understand undocumented events …
Try and type « deep learning » and « Field Service » as keywords in your browser : you will probably return articles rather dealing with machine learning. This for the simple reason that most of FSM softwares are using this learning technology.
Machine learning works well enough to make an automatic diagnosis on identified problems : the computer is fed with problem / solution pairs based on experience, and when a malfunction occurs the system strives to link it to the right pair. Machine learning relies on clustering and decision trees (questionnaires for example).
But when symptoms resemble nothing known, or when documented solutions are uneffective, it becomes necessary to search further, or, in this case, deeper. This means for example being able to understand that, even if the malfunction looks like case X, it also has in common with case Y a minor symptom which is indeed the crux of the problem.
It can be done using a deep learning system with several layers of neural networks. The idea is not to cluster data sets but to thoroughly understand those data in order to « extract » knowledge from them and then apply this knowledge automatically to dissimilar events.
For the outcome of this process to be relevant, it is necessary to solve a second difficulty in the equation : data sets with the highest knowledge content are generally texts. Hence the need for semantic analysis.
New methods for natural language processing, such as vectorization or word embedding enable to link words which belong to different semantic universes by analyzing the context in which they are used.
With the association of deep learning and semantic analysis, automatic diagnosis is taking a giant leap forward.
… and improve your KPIs !
A software which is able to suggest solutions that are not yet documented will help increase your first-time fix rate, and consequently your customer satisfaction.
If this software is, in addition, capable of efficiently modeling problem-solving processes by integrating in real-time data coming from IoT sensors, your teams will also save tremendous time on each operation. Productivity will peak.
More, the sequence history of each maintenance operation will be registered in the knowledge base. It will add up to all experience feedbacks, coming from all over the world. PICC Software will be able to network all these experience feedbacks, whatever the language or vocabulary specificities used, to offer an unprecedented detail level in root causes analysis.
Central services will hence be in a position to reveal causal links they would never have imagined.
Willing to know more ?
Feel free to ask us for a demo.