PLM Process Evolution - From Semantics to Autonomous Agents
Sumesh Sadasivan posted on December 19, 2013 | | 9185 views

The Process vs. Data debate in PLM has been debated over and over. You can read the discussions on Oleg's blog or watch a debate by Jim Brown and Chad Jackson. I am not going to get into the merits of process vs. data. My point in this post is about how processes will evolve from a PLM perspective.

The current state of Business Process Management within PLM

Organizations are driven by time-to-market pressures, varying market dynamics and regulatory changes all of which impact their organizational processes and data requirements. Most of the major PLM Systems have Business Process Management (BPM) solutions that are efficient in execution, but are not easily reconfigurable in response to changes.

These BPM tools also tend to lack automation capabilities, instead needing human intervention. This lack of automation represents a real opportunity for future development, since a good percentage of decision making scenarios could be deduced from historical data. For example, a Change Management process can be influenced by factors like how much a change would cost. That data might prompt the system to look at an alternative part to bring down the cost. Or, an automated system might dig deep into the archives to pull out a design that represents a more cost effective solution.

In short, we can summarize the main shortcomings of BPM within PLM as:

  1. Higher cost of change
  2. Lack of automation

What a “better” BPM System would look like

To address these shortcomings, what we need is autonomous system that can take decisions on its own as well as adapt to changing industry practices; essentially an ability for dynamic workflow co-ordination/re-configuration.

Autonomous systems are composed of Agents that will have access to interpretable information allowing them to make intelligent decisions. That would require the systems to have a good semantic knowledge base. A good semantic knowledge repository has to be backed by a well-defined domain ontology. Over time, the ontology will evolve by inferring new rules thereby enabling even better decision making.

We are already seeing a focus shift from syntactic to semantic data in the PLM industry. Autodesk PLM360 has a built-in semantic engine, DS has Exalead as part of its V6 platform while Oracle Agile has the EDQP integration. Once these solutions reach a certain maturity level, and a domain ontology becomes the de-facto model, we should see a general shift in favor of Autonomous agents. These are not yet main-stream in enterprises, but are predominantly in research/prototype now.

I have to add that social media can give us insights into the future of this type of innovation. For example, Google recently patented a software that can manage your social media account for you. This isn't something new, as Google has been using variants of this "Intelligent Inference" capability in its products. However what makes this different is that we are seeing Autonomous agents that are sufficiently mature to take over decision making in the absence of humans.


About the Author:
Sumesh Sadasivan is a PLM Consultant, software developer and technology enthusiast. Sumesh holds a Bachelor's Degree in Computer Science and is working as a PLM Consultant with Tata Consultancy Services. He has been part of the software industry for more than 13 years and has vast experience in implementing PLM solutions across industry domains and packages.

Recommended For You