In a new episode of ‘Startup Stories’, UpsilonIT talks to Tomer Shamir, the Co-founder, and COO of AiDock, a company providing AI solutions for paperwork automation. From Tomer’s story, you will learn how they are delivering value through artificial intelligence to the supply chain paperwork and why it is important for the logistics industry.
Today we're talking with Tomer Shamir, who is the co-founder and COO of AiDock. Could you tell me a bit more about your background and how you got into the logistics and transportation industry?
So, my first experience in logistics was about 15 years ago. I was a customs clerk for one of the export departments in a freight forwarder company in Israel. Since then, I have moved into many operational roles in the company. My last role was as the VP of strategic customers in a full global freight forwarding company. I was mainly dealing with global shipping and logistics. It's always been a regional issue that you can move a shipment from Japan to Germany in a few hours, but it may take a few days to get customs cleared. That was always a big pain and a problem that we always addressed and tried to be ready to. We wanted to minimize the time that we need to wait for a customs declaration.
About three years ago, I met my partners, Eddie and Sean. They wanted to automate those kinds of paperwork processes. As for me, a person who started by doing that paperwork and finished managing those processes, it seemed a bit unreal, but I felt that if this is real, I want to be a part of it. That is how I got into AiDock as the domain expert of customs with my two partners. One of them is a longtime entrepreneur and experienced software engineer. And the other one is an AI leader here in Israel, and he's actually the one that created the technologies that we build our solutions on.
Let’s talk about AI more. We see that the current hype around AI in logistics is mostly connected with emerging technologies like price predictions, self-driving trucks, drones, delivery robots, and so on. Can you explain how AI has changed paperwork-related tasks on the AiDock example? What is AiDock doing?
Okay, so before we dive in, I'd like to give you an overview of AI. AI is a great buzzword, but actually, it means nothing. It's like saying I have the Internet. Well, that's cool, but what are you doing with the Internet? You need to transform the technology into a valuable product or create a business value that helps you do more, better and faster. A very easy example is Tesla, the autonomous driver. So, the autonomous driver is AI-based. It enables the driver to predict the next right decision, and it is up to the driver to decide whether he or she wants the car to make the decision for them. It's only a matter of trust eventually.
So the first time using an autonomous driver will be very scary, and you probably will not move your hands off the wheel, but ongoing, when you get the trust that it makes the right decisions, you will probably let the car take them for you. And it correlates with what we're doing with paperwork in many touchpoints. Paperwork in the supply chain is critical so that you can have a typing mistake with a shipment and a package can be delivered to a different destination, or you can have a typing mistake with customs, which can cause a lot of problems. From my experience, when you have customs problems, you eventually lose the customer because no one wants to deal with customs issues. That is actually what our system is doing with paperwork. It gets a paper, understands its type, and helps to decide how to process it. When you combine the same as Tesla, an autonomous driver with a human driver, you have no accidents, so that's 100% success. It allows us to lead our customers to 100% accuracy and maximum efficiency.
Another pillar of paperwork is classification. It is a very hard and time-consuming task and has many implications on financials. If you're using the wrong HS code, you may be over or underpaying duties, follow the wrong regulation, and the customs will get you fined for that. So Cody, our virtual HS code classifier, is the one to help in that process to predict the right HS code for every item, or flag risks encountered by previous declarations. When our customers use all the solutions that we offer, they're actually creating maximum controllable accuracy for HS codes. They significantly increase the accuracy in all over the customs declaration, reducing the risks, and a very important key here is reserving the knowledge. So, when you have virtual employees, every task you are doing is knowledge. When virtual employees gain that knowledge, they can analyze and predict better decisions in the future.
Where do you get the initial data to train your algorithm from? Is it publicly available data from customs, or is it your clients' data?
It's like a joint venture between public data and customers’ data. You may hire someone with a degree in customs or something that helps him better understand the processes, but it is not enough. When you hire someone, you need to train them on your actual data and your real customers, and on your existing process. Even though most companies have similar processes, you still need to match some differences and classification patterns. So, when you combine the basic knowledge of classifying with the customer data, the virtual employee can learn the pattern and the type of process that they're using, that's where you get a maximum value from the product.
How do you deal with different types of documents that vary from country to country?
Every country has its own modifications, but it's not a new process. There is the basic part, which is the general customs process and knowledge, and there is the customization of about 20% to 30%, where you need to make some changes in the product to make it fit the designated country.
That is basically what we're doing, we take a virtual employee that knows how to do the job, but we still need to teach how to do it better locally. And this is one of the key parts of AI. training is not being done on the customer's payroll, but training is done before you hire the employee. So you hire the virtual employee only if you are able to teach him what he needs to learn and he passes the test successfully. Only after this, companies start using the service.
How do companies react to the idea of implementing virtual employees and AI into their working processes? What are the main things they need to understand? What are the biggest challenges for companies to start using AI in their daily work?
We have customers from various industries, and each one has its own use case and challenges. But I believe that the biggest challenge is trust, and the rest will be easy. Once we have the trust and the understanding of what technology can do to your company and how it may affect the bottom line eventually. It is critical for every supply chain service provider to have those AI capabilities for the paperwork, which is the biggest bottleneck in our industry.
How fast are you able to see the business value of AI implementation and show results that the system is actually working and saving you a lot of time and money?
It depends on the customer and his pace. From our experience, working with several customers by today, takes about 90 days, mainly due to systems integration. Customers have local and sometimes very old systems, and we have the new ones here at AiDock. So combining them sometimes takes a bit more time, but as I said, once you gain the trust and understand the value, that's definitely something that will not hold you back from partnering with us.
While robots and bots are improving business outcomes, there's a growing concern about their potential to eliminate jobs. How do you help your customers overcome this fear? And how, in general, do workers react when they start seeing the virtual employees and the AI in work?
In a general overview, technology never eliminated jobs. It never happened in the past. The world is evolving, and things are changing. The fact that a company will have virtual employees does not mean that it will create unemployment problems. On the contrary, it will help their employees deal only with the top 20% of the issues that need human intervention. It will clear their time to do other tasks much more accurately to increase the company's efficiency and accuracy and eventually deliver better results.
In addition, there is a general problem of hiring people for customs. Usually, companies have the problem of not getting things done fast enough. So actually, our technologies can help the companies deal better with the current work and deliver faster with the current volume they have and give them the infrastructure to grow exponentially without hiring more employees. This is the part where virtual employees come in to help the company grow without the growing pains.
What are the three most challenging things you have faced since the company started working, and how did you solve them?
COVID-19 has changed a lot. During the pandemic, people learned more about what AI can bring to the company and its advantages. But I still think that the only challenge is to create trust so that companies can trust the technology and understand its value. The rest is actually not a challenge because everything can be solved. Once you advise a customer that they can grow with your solution, make more money, and be more efficient and accurate, and they're also reducing the risks so other issues can be overcome relatively easily.
You briefly mentioned that COVID has changed a lot in the way companies conduct their business. How has COVID-19 affected the logistics industry, customs, and technology in general?
I think COVID, first of all, proved that logistics and the Internet are one of the most important parts we have in life. Without the Internet, we wouldn't be able to order the masks, and without logistics, we would not be able to get these masks. But pandemic has also revealed a lot of problems. For example, customs agents that use traditional, not automated processes and methods understood that they cannot do their work anymore because they cannot come to work right now, they need to work from home. They need to use technology to help them overcome this issue. And I can tell you that last year, once COVID started to affect the world, a few of our customers told us that the innovation budget is over because now they need to support the company during this pandemic crisis. But in 2021, the same customers got back to investing in digital innovations. So now they are assigning a lot of funds for digitization, efficiency, and technology. If anything like that happens again, it will not harm the business, and they can keep up working as they worked from the office, but now they can do it remotely with the assistance of virtual employees to overcome any crisis on the way.
What trends in technology and logistics are you expecting to see in the next three, five years?
Technology will change how we do stuff in logistics, starting from autonomous trucks for the pickup and last-mile delivery by drones, the rise of e-commerce, and many challenges over customs that will make companies use technology. For example, the customs process that you have to declare on every item valued over 22 Euros with a full HS code. If down the line governments will have better statistics on what's been imported, they can collect the tax revenue that they're missing today and generally create a better environment for everyone. As a result, they can make sure that every item that enters any country is enforced with the right regulation and pay the correct taxes so that everyone can enjoy the benefits out of it. It is one quick example as you know about 200 million parcels are moving internationally every day. You don't expect to have clerks to examine or to advise on an HS code for every item, just impossible. You also cannot expect the customs authorities to check every parcel because it's also impossible.
On average, only 0.01% of the parcels are being checked, and it is a problem because a lot of things are going under the radar, which should not. As a result, they lose revenue. So, I think that the governments and the organizations will increase the regulation and enforce them because technology allows that now. Back in the days, to check, 1000 HS codes would have taken you 1000 minutes. Today it can take you a second. Now you're able to have much better control, accuracy, and efficiency in that perspective of customs. In the coming years, I think we will see more regulations coming in and technologies taking a bigger part in the supply chain.
Do I understand right that the government also needs to implement technologies in order to be efficient and keep the same pace as the logistics companies?
Usually, customs are not at the same pace as the industry. What they would choose is to ask the companies to enforce the regulation. So the companies would use the technology, and the customs will monitor the customs agents using technology. They have control over the accuracy, and they know what's coming in, what's coming out. It will be much faster and better in that way rather than having the customs examine everything, even though it is also a possibility. I know that some customs authorities are already using technology to examine items and shipments better. So they make sure that they comply with the regulation.
How is the AiDock solution managed? Could you tell us more about the technologies used and how it works?
We've created here at AiDock three pillars of deployment that would fit different types of customers. Each one has a business value to the customers, so they can really gain that value by using the system. The first one is the web platform that accompanies our virtual employees. Human clerks can do the job on the system by themselves and use a regular customs declaration platform, and with a click of a button, they can also have a virtual employee to help them. For example, if you're a customs clerk and you're usually handling 10 customs files a day, and today you have 30, then with a click, you can switch to a virtual employee, and it will help you to do all the work. You only pay for a finished work and not a subscription plan.
The second one is the API. It's a cloud solution designed for companies with their own IT department, and they can connect their systems to our API. So, down the line, the work of our virtual employees is seamless. Companies send a request from their system to ours, and it is doing the processing and sends back the answer, so eventually, clerks have the same system; nothing has changed other than the blank fields they had previously to fill up in the customs declaration are already filled with data. So it's a very easy change for companies that can do this kind of integration.
The last one is an On-Premise solution, which we have because some of our customers have cyber requirements that do not allow us to use the cloud. So we also have a solution for them.
Thank you, Tomer, for sharing your experience. And thank you for this interesting conversation!