Last week I wrote about a trade show conversation with someone stuck in year two of an automation implementation – they had a modest feature list, blown budget and a team beyond frustrated. The point I made was that the feature list was never the problem. The implementation was.
I wanted to extend those thoughts this week, because the tools to solve that problem are evolving remarkably fast.
There’s a lot of focus on AI in print on demand right now — automating customer service, quote generation, artwork creation, prepress. All legitimate applications. But there’s a potentially more impactful use case that gets far less attention: using AI to solve the single biggest bottleneck in moving to factory automation……
……. getting live.
As I covered last week, one of the core challenges in automating your production is having a solid foundation in your product codes and product metadata, your imposition library, your workflow rules – how you actually run the factory – and your shipping logic. The reality is that most businesses don’t have that foundation data beautifully organized. It’s evolved organically over years, with edge cases and exceptions layered on top of edge cases and exceptions.
That (messy) reality has traditionally meant one of two things: either you spend months cleaning and mapping your data before you can go live, or you go live with gaps and spend the next year firefighting.
What’s changed is that you can now train AI systems to take the data you’ve actually got and infer your internal product codes from it. The AI learns to map those to your internal processes, and where there are inconsistencies – and there always are – it learns the patterns in your data rather than requiring you to fix every exception manually upfront.
What that means in practice is that instead of a big-bang, heavily resourced go-live effort, you can take an iterative, learning-based approach. The system gets smarter as it sees more of your real orders. It’s an organic approach to going live, not a forced one.
The second part of speeding implementation is creating all the print-finishing workflow impositions that a full automation solution needs. We’ve been applying AI to that problem as well. You can take an existing imposition – however it was built – feed it to an AI-driven imposition engine, and have it automatically convert that into a rule-and-recipe-based imposition that integrates with your core automation platform.
So to answer the question I posed last week about challenging your automation supplier on implementation timeframes: the answer should be fast. Not because corners are being cut, but because the latest technologies are fundamentally changing what’s possible in productising your products and impositions.
If your current or prospective automation partner is still quoting big (or expensive) implementations, ask why. The tools exist to do this differently now.
If you’d like insight into how this works in practice, reach out – we’ve got a lot of experience in this space.
