What To Build: Peter Puchwein (KNAPP)

Conversation with Peter Puchwein, Vice President Innovation of R&D @ KNAPP. We talk about KNAPP’s process for selecting the right types of customers to pair with startups to ensure a smooth transition from pilot to production, and specific examples of existing problems within fashion, e-commerce, pharmaceuticals and general warehousing that require deep AI & robotics talent to solve.

1. Tell us about yourself, where you currently work, and your path on getting there.

I have been working at KNAPP AG for 15 years. The KNAPP Group is a global provider of intelligent automation solutions for intra-logistics, specializing in healthcare, fashion/apparel, general retail, food retail, industry and service. KNAPP has more than 4,200 employees across 38 locations worldwide, with an annual revenue of around EUR710 million. The majority of our business is in Europe and North America.

I come from a software development and mechanical engineering background and have been responsible for my firm’s research and development division since 2011. Our R&D division is comprised of ~300 software and mechatronics employees who design and develop the core products.

2. Tell us about your role and what your mandate is and how this specifically relates to working with startups?

We are always searching for new ideas and technologies whether coming from university labs or startups that can help us with increasing automation across our company.

Automation needs are different in the USA versus in Europe. In Europe, the level of automation is higher because we rarely have people doing manual fulfillment work here anymore as labor costs are very expensive; therefore, our European customers need higher levels of automation because they can’t employ people for this type of work. But it’s different in the USA as there’s a higher likelihood that you will be able to find more people who are willing to do this type of work although it’s becoming more similar to Europe.

We are looking for startups mostly in the robotics field and we’ve thus far seen hundreds of companies over the last few years focusing on AI & robotics. There’s a big hype around AI and robotics but we’ve only found a few startups who understand the market and market needs and are not just another webpage ending with dot AI.

Currently we’re working with ten startups: we always start with a proof-of-concept (PoC) or prototyping stage with the hope of being able to make it production-worthy for our end customers. We sometimes invest in these startups but that’s really not our main focus — our goal is to be able to integrate their technologies into our product offerings for our customers.

3. What are some of the interesting types of projects that you’re currently doing with startups?

Unfortunately, I’m not at liberty to talk too much in detail about our projects but the focus is on AI and Robotics.

4. What number of these projects move into production? By what criteria? One of the challenges we see startups facing is how to move a customer from pilot to production.

From our end, if the prototyping stage goes well, we can then move into production with the startup. Sometimes the prototyping stage can be as short as two weeks. If the technology is working, we then show it to potential customers who we think might have interest, and then quickly move into production. We rely on our customers because while we have laboratory testing facilities at KNAPP, we don’t have all of the hundreds of thousands of items (i.e. products) that are important for the startups and so we’ll need the customers’ support.

For newer technologies, we elect to partner with existing customers who have worked with us for a long time and who trust us. We’re very honest with them — we tell them we have a new technology and ask if they’d like to put it into production together with us and maybe share the risk. We are also careful to plan implementations at the appropriate times — for example, a time when business is not at a seasonal high.

One large criteria for moving a prototype into production is in the level of support that a startup can provide to customers. Our customers need 24/7 support every day of the year because any stop in production means they’ll lose a lot of money as we provide critical production technology. For example, what happens if your technology breaks down? What’s the Plan B and Plan C? We’ll need to think through potential solutions together with the customers; the customers are taking a risk but it is a calculated risk. We spend a lot of time visiting startups, seeing what they’re capable of (vs. just seeing online video demos of their technology), and discussing all the details of our customers’ needs with them so that we can make the best decisions as to who to partner with.

5. What are the major challenges in your industry these days, and specifically ones that you think can be addressed by the right type of AI and or robotics application? Can you give some detailed examples?

There are so many examples of challenges that can be addressed using AI and robotics, which I really do think is the next big wave. I’ll just name a few examples I see across different industries.

In the fashion industry, if you’ve bought shoes or apparel online, it takes a lot of manual work in the warehouse to put what you’ve bought into shipping boxes. We’re hoping AI and robotics can help automate more of this process; automation can be especially hard for the fashion industry because apparel is often kept inside polybags (a type of small plastic bag) and it’s not possible to grasp this with existing technologies right now. Imagine a big box with hundreds of these polybags inside, with only humans being able to detect and grasp them right now. We need more automated solutions here.

In the general e-commerce space, people are buying a huge variety of items online and, again, we don’t have a good enough solution for picking and grasping and handling all of these different types of objects when they’re kept together in one box. Let’s take the returns process, which can be upwards of ~50% of goods sold online depending on the merchant. We need an automated solution that can handle boxes with many different returned items inside — the system would need to know which items are in the box, how to pick up each item accurately and efficiently, and where and how to put the items back into storage. This is easier to do if the box is filled only with the same item type but that’s not how returns work.

In the pharmaceutical wholesale industry in the USA and Europe, regulations are becoming more strict regarding what’s permissible (i.e., not dangerous) for consumers and hence there’s a greater need to be able to automatically read labels off of drug bottles and packaging. For example, picking out each item from a box of 25 products and scanning the 2D code (provides information on ingredients and expiration date) of each product to see if it meets regulatory requirements. Moreover, the items need to be scanned at each step of the supply chain process for safety reasons.

In the warehousing space generally, we’re building fully automated storage and retrieval systems (ASRS) with millions of storage locations located inside these racks. In the future, we think there will be micro-fulfillment centers in city centers as well (vs. only in remote areas where majority of current warehouses are located). For example, you might have micro-fulfillment centers in grocery stores that allow you to pick up your online grocery orders at the counter when you come into the store versus your having to walk around the store to pick up items one by one. We think the industry is heading in the direction of small local solutions — these solutions need to be tied into the point of sale system and need to be easy to setup, maintain, etc.

6. What type of startup would you be most excited to see?

We’ve found that many startups are good at PR and marketing but less focused on providing real solutions. Sometimes technical founders don’t understand what an industry needs — this seems to be the case for 90% of the startups we’ve engaged with; in such cases, it’s critical to have experienced people who have industry experience involved in some way.

7. What should startups know about your industry before going in? What nuances or details about the industry are not so apparent from someone looking in?

Startups need to find partners in our industry; they need to find potential customers, visit them and see exactly what problems they’re facing. We see startups going to market too late — they’ve been working on a solution for six to twelve months and only afterwards realize customers don’t want what they’ve built. We help our startup partners meet with our customers in day-long workshops so that they can fully understand the challenges they should be building solutions towards.

Our focus is on five big market segments: fashion, general retail and e-commerce (online), food, healthcare (wholesale), industrial solutions for automotive and production. You can build a solution specific to each of these different industries or, better yet, find a way to build an underlying core systems that cuts across all these verticals and tailor what needs to be differentiated for each one (ex. underlying core system with different grippers and end effectors for picking and grasping).

8. Lastly, any recommended resources / reading (ex. Industry conferences, publications, experts to follow, etc.) for startups looking to build in your space?

In some cases, I think It’s enough to do online research on markets and customers. At least 80% of the relevant information on industries is online but of course you need to take the time to read through the information carefully. If you find potential customers online go talk to them. The most important thing is to talk to the end customers and visit their warehouses. Conferences can be good for that as well.

Another avenue is doing patent research in order to see who else has done something similar and potentially contact other people working on similar technologies.

We’ve met the startups we currently work with in different ways — some came in as referrals, some we met at conferences, some are cold inbound, etc.

If you’re going to send cold inbound messages please note that your end customers are likely getting a lot of such notes and we don’t have the time to read through pages and pages of information so it’s good to send a clear concise blurb on what you’re working on.

Click here to learn from more experts for our What To Build Series.