How to improve PLM implementations with artificial intelligence and machine learning

How to improve PLM implementations with artificial intelligence and machine learning


PLM implementations are usually taking long time, cost money and resources. Manufacturing companies are blaming vendors for overselling and excessive marketing. Software vendors are blaming customers for inertia and organizational change management problems. All together agreed that PLM implementations is something that slowing down PLM technology adoption within companies.

Typical PLM implementation is a service project, which combines specific customer requirements, PLM software product and  customization (let’s not talk about configuration vs customization for the moment). An interesting part is that implementation part is often repeating from customer to customer. Customer requirements and management overhead are main drivers for inefficiency. PLM service companies are looking for applications engineers to work togther with project managers and domain experts to solve a problem.  Combined with high demand for skilled engineers it sounds like unsolvable problem.

Software eats Software?

Software is eating the world. We know that. But here is a new formula – Software eats Software. Techcrunch article  speaks about Gigster – the company aimes to solve a problem of high demand on development software and applications. It suggest a fundamental shift in how software will be built. Think about it as a “software development company without engineers”. The core element of Gigster is a combination of skilled engineers (pro-lancers) and artificial inteligence engine able to convert customer requirement into development plan.

The lauded venture firm was impressed with Gigster’s artificial intelligence engine. It converts a client’s product proposal into a development plan, and helps Gigster’s army of remote developers plug in pre-made code blocks to efficiently build the app. Built by co-founder and CTO Debo Olaosebikan, Gigster’s AI perfectly fits Marc Andreessen’s investment thesis that “software is eating the world“, and Andreessen partner Chris Dixon’s thoughts about “software eating software development“.

On the surface you might think about it as a glorified outsourcing development company. But, in fact, it is not that. The core of Gigster is development of artificial intelligence engine that allows you to re-use code between projects. The following Business Insider article describes it as Gigsters’ secret weapon.

There are plenty of other freelance marketplaces out there, for programming and for other services, sure. But Gigster has a secret weapon, in the form of a “smart platform” built out by CTO and co-founder Debo Olaosebikan, Dickey says. Behind the scenes, Gigster uses machine learning technology to get smarter over time. Basically, it can figure out the things that are similar about different customer projects. That’s useful not only to figure out the best developer for the job, but also to suggest using code that helped complete similar projects in the past. It results in a shorter time to completion.

From Gigster to PLMster?

Do you think PLMster is another fancy buzzword? Maybe… or maybe not. The connection is probably not obvious but here is what made me think about PLMster idea. On average, PLM services are representing about 50% of every PLM implementation. The technology and PLM software is important, but often the key component is skilled project manager and service organization that capable to convert customer requirements into successful PLM implementation.

Usually manufacturing companies are buying projects from established service providers such as IBM, Accenture and others. These companies are getting tons of PLM projects because they’re well established. But most of these contracts have premium price tag. You basically pay for “insurance” and “brand”.

At the same time, the failure of PLM implementation projects is usually not in the technology, but in the ability to capture customer requirements and combine it with the most efficient technology to get job done. This is a point where I can see an analogy between Gigster ideas and challenges of PLM implementations.

Service companies are trying to apply ideas of how to re-use existing practices for multiple implementations. Recent Razorleaf announcement introduced us to Clover integration platform. It might sound as a middleware, but in fact I don’t see it that way. In my view, Clover is a software code which represents implementation best practices used in many Razorleaf projects. Navigate to Razorleaf blog to read about Colver for Autodesk PLM360 and how it can be used on top of existing middleware – Jitterbit. The secret sauce is how to stitch things together and it is usually hard to do without code.

What is my conclusion? PLM implementations is a point of failure of most of PLM projects these days. It represents the most inefficient elements of PLM vision and it slows down PLM adoption rate. OOTB (Out of the box) deployments, flexibility, easy configurations, cloud, open source… what else? PLM industry was trying to come with reasonable solution to speed up PLM implementations, but unfortunately, without visible success and traction. PLM implementations are still complex and taking time. Artificial intelligence engine for PLM implementations? Sounds crazy? These are just my thoughts…

Best, Oleg


Share This Post