Automatic logistics planning for a large craft business
“How physicists solve problems: You have to come up with that first!”
Mark Camporelli, Brickwork Ventures GmbH, Switzerland
Project key data
2 people
6 months
Intelligent planning application
Physical models, algorithms
Java
Docker
Why we like showing this case
‘Craftsmanship’ evokes things that are practical, tangible and often close to us. Like a beautiful kitchen, for example, in which you can celebrate a good start into the day. ‘Algorithms’ and ‘physical models’, that sounds abstract, but allowed us to help a large kitchen manufacturer in a very practical way. They were able to leave tedious manual scheduling behind and to make beautiful kitchens for more people in less time with automated planning.
Task and solution
Brickwork Ventures GmbH had already developed a feature-rich digital planning tool: "Optibrick". A chief user of Optibrick, a large Swiss kitchen manufacturer, faces complex, nested logistics chains and challenging management of many different capacities. Planning, if performed manually, is time-consuming and error-prone. Optibrick was therefore to be supplemented with a planning engine.
The algorithms of such a planning engine have high mathematical and logical requirements; therefore Brickwork - itself a competent full stack development company - relied on our experience in algorithms, mathematical and physical models for the development of the new core ingredient to their platform.
Where can physical models help?
Physical models often add value where you might not think of physics at first, but where precise prediction and simulation are required in operational management or finance: Capacity and resource planning, process optimization, risk minimization, decision management and much more.
They are a versatile and effective toolbox, provided you know how to use them. What we particularly like about physical models:
They are easier to interpret and understand than many other statistical or machine learning methods. The "black box" that is often inherent in machine learning and creates distrust against "artificial intelligence" is not present here; instead, we gain a good understanding of the underlying processes. Physical models also provide excellent opportunities for extrapolation beyond the range of given data, allowing us to predict future trends and phenomena.
Where does automated planning pay off?
The leap to automated planning is often smaller and less expensive than you might think, but can significantly increase cost efficiency and customer satisfaction.
When planning processes are complex and involve many variables, dependencies and constraints, automated solutions quickly outperform even years of experience in "human" planning because they objectively process large amounts of data and take complex relationships into account to generate optimized plans. They enable faster response to changing environments and reduce human error, while allowing planning in scenarios. By analyzing planning data, companies can make informed decisions, optimize their business strategies, and gain competitive advantage.
→ Questions or thoughts about physical models or automated planning? Dr. Markus Abel is looking forward to hearing from you.
Interview with Marcel Camporelli
Greetings Marcel. Please tell me in your words what was the challenge for Brickwork at that time.
Our customer, a large kitchen builder, wanted an automated, intelligent planning tool. Our product, Optibrick, was morphing more and more into a full online scheduling solution. And that's where you had to build a planning engine on top of the base system of Optibrick that would create automated planning from constraints like capacities, customer satisfaction, and so on.
And that's when you needed Ambrosys? You are a software company yourselves.
We are very strong in implementing full-stack software systems ourselves, but you need a focus, don’t you? We didn't want to become more deeply involved in the algorithms of automation, because that's where it gets mathematical-logical pretty quickly, where you have to get really deep, right? We then decided to contract it out to Ambrosys - the Markus-and-Markus team! They have already solved many more such problems than we have, problems that involve mathematical optimization algorithms. For us, this would have been absolute frontier territory. Yes we had a proof of concept, but it wasn't really mature. And that's where Ambrosys really helped us finalize and get it to a level that would work reliably.
So for what purposes would you recommend Ambrosys? Or for what kind of clients?
I guess there's all different types of clients they’d help out, right? So I can answer it more for myself, I don't want to say that they don't do other things in a swell way. But for us it was really mathematical optimization problems, complex optimization problems. What is perhaps a specialty at Ambrosys: their professional background. There are a lot of physicists and mathematicians milling around there. You can tell. For the problems we had, that was awesome, a great quality.
How does a physicist deal with problems?
I'd say: They have a powerful toolbox of mathematical, partly also logical tools to solve problems. And then they also have that knack to understand, when it comes to completely new problems: What tools could I use for this problem? And that would be tools no one else might even think of. Then they sometimes use solutions that come from completely different areas, from thermodynamics or whatever, but it turns out those tools can solve your very capacity planning problem. A normal person wouldn't have come up with that in a million years, would he! And such is a physicist’s mind.
Apart from the physicists' toolbox, how does it feel to work with Ambrosys?
We have found the team to be very competent. Also open and flexible. Nice, competent, flexible. I also found all the people I dealt with very, very pleasant to deal with. So I actually came out of the story very positive. I will definitely go with them again with tasks like that.
Thank you very much, Marcel. That's it. That's how efficient the Swiss are!
That’s for sure a thing we have in common with our friends in the “big Canton”. Ciao!