the factory that runs itself

Publishers are increasingly experimenting with small print runs and on-demand printing. The critical issue is time of delivery here. Isn’t that a problem for their logistics – a book that takes four days to reach the bookstore?

That’s true, but our software gives us direct control over that. What we’re actually achieving for many of our customers – often mass-customisation businesses or online printers – is a dramatic reduction in lead time. We’ve had customers who previously needed three weeks for their average product – such as food packaging. They’ve cut their delivery time down to 48 hours. What was missing in the book trade until recently with the “Book of One” concept wasn’t the production logic but the willingness of the entire supply chain to adapt to it. That is changing rapidly; we are currently providing the software for newly emerging factories in Germany, Europe and North America that are entirely geared toward the production and distribution of “Book of One” products.

What is the fundamental difference between your approach and the way a traditional print shop operates today?

The key difference is this: We don’t work on a job-by-job basis but rather on a product-by-product basis. A traditional print shop calculates each job individually, determines the production steps during prepress and then follows this plan to the letter. Ultimately, this is a craft-based approach – artisanal but not scalable. Our approach is truly industrial: You define a product or a product group – such as a book with attributes determined dynamically at the time of order, like format, features, substrate and colour scheme – and production intentions are stored for this product: printing, cutting, folding, binding and packaging. Then an intelligent planning algorithm analyses the entire job pool and decides which jobs run on which machines in what order – with the goal of meeting all delivery deadlines at the lowest possible cost.

That sounds like fully automated production.

We call it the Smart Factory. The product finds its own way through the factory independently. Of course, there are still people who load paper, operate forklifts and intervene physically. But control of the factory no longer lies with the production scheduler – it lies with the software. Zero touchpoints as the ideal: The order comes in, is produced, picked, shipped and, if desired, automatically invoiced.

Publishers are generally considered to be somewhat resistant to automation. The manufacturer calling the printer to negotiate delivery times and terms – that’s a classic image of the industry.

That’s true, but it’s not an obstacle. We accept orders from any channel. Whether a phone call was made beforehand or not – that’s completely irrelevant to us, but it probably doesn’t make sense for Book of One. What matters is the moment the order is finalised: Once the specifications, quantities and delivery parameters are set, the system takes over. And up to a certain point in the process, orders can still be changed – you simply start a new planning cycle. What comes after that runs fully automatically.

Why hasn’t mass customisation really taken off in the publishing industry yet?

I can’t give you a definitive answer to that, because we’re the implementers – not the strategists. When marketing managers at publishing houses come up with good ideas for personalised products, we make sure they can be produced at scale. But what I can say is this: Mass customisation can only become a mass phenomenon if it’s affordable for the end customer. If a photo book cost 500 euros, it wouldn’t be a mass-market product. If it costs 24.99 or 39.99 euros, people will buy it – and that’s only possible if the process is fully automated. There are certainly examples that work: there are startups that produce children’s books featuring the child’s name and a personalised storyline. It’s a huge hit. Just not with traditional publishers.

What technical standards do you use for machine integration?

JDF/JMF is the industry standard – unpopular but de facto established. The problem is that it’s documented in a 1,000-page manual and allows for many different interpretations, which regularly drives developers to despair. The original hope that an Indigo could be controlled just like a Canon or a Heidelberg has not been fully realised. But at least there are no longer any completely proprietary machine control systems. And where standards end, we come into play.

How do you tackle the issue of complexity with only about 50 employees?

It’s quite a challenge, yes. We develop the software ourselves and have lean and highly efficient processes. This allows us to build software that can do more than some products from large corporations. We built Symphony to be very generic – with no-code editors that prevent IT development monopolies. Large printing companies with their own development departments often have hard-coded systems where every small change requires a huge amount of effort. That’s our structural advantage. And in addition to software development, we offer printing companies comprehensive assistance with implementing the Smart Factory if desired, because our editors are no-code but certainly not no-brain.

What role does AI play in your system?

Our planning algorithm is based on AI methods – specifically, on a system that does not rely on large language models. It’s not about saying that a job pool would likely run best with this machine. It’s about calculating that precisely. A heuristic solution would take hours to run through all the combinations. Our AI algorithm works with reward systems, quickly discards what is obviously inefficient and delivers a result in five minutes instead of five hours. This is an AI you can’t talk to – it simply does its job in the background. The AI you can talk to is the classic agent that helps users work with the system more easily. Still not “no-brain,” but perhaps “less-brain.”

Are there industries outside of traditional book printing that implement your approach particularly consistently?

Online printers were our historical core. That was formative for us. But now we have customers in the confectionery industry, among infrared heating manufacturers who print custom designs on their products, at packaging printers, at manual printers for machine manufacturers and of course in the field of mass customisation, such as photo book and wall decor printers. Anywhere individualisation meets volume, you need the same logic. This is no longer just about printing – it’s about production logic.