When you order a package from Amazon and Wish, it almost feels like magic. With the click of a button your consumerised needs are met in a matter of days. However, we all know that this is hardly magic, rather it is the work of some of our most complex machine learning systems and warehousing technologies. Let’s dive into some of these technologies and firms which are making this process seem even more magical!
Written by: Tanay Sonawane, Jonathan Ouyang, Filip Vrábel
Robotics in Warehousing
As AI and automated machinery becomes more sophisticated, it is no surprise that these technologies should spill-over into the warehousing industry. In recent years, the dramatic improvements in robotics technology has allowed the implementation of automated systems to manage complicated processes, or perform dangerous tasks in the warehouse. Indeed, with warehouse automation, you can expect to see huge uplifts in efficiency in the management of warehouses and inventory.
Once you’ve placed the order of that human-sized teddy bear you so want (let’s be honest, we all want one), the teddy bear, let’s name him Bob, will be picked up by an Automated Guided Vehicle (AGV). The AGV, resembling a forklift, travels along a predetermined path in the warehouse, using sensors and magnetic strips to navigate, transporting the teddy bear onto the next stage.
Once Bob is at a transfer point, an Autonomous Mobile Robot (AMR) takes over. Now, an AMR is a more sophisticated version of the AGV, where it does not rely on fixed tracks, but uses advanced sensors and SLAM (simultaneous localisation and mapping) technology to move through the warehouse. The AMR will skilfully weave our beloved Bob onto the aisles, where it will be picked up by the Automated Storage and Retrieval System (AS/RS).
The high-rise robotic library of goods, shuttles and tracks that is the AS/RS will place Bob in his designated spot, maximising space efficiency, until it’s time for Bob to be dispatched and to meet you!
Once your order is processed, a Collaborative Robot (Cobot) will assist warehouse workers by picking up Bob from its storage location and acting as a storage trolley. It will follow the workers through the aisles, reducing their physical strain and speeding up the picking process.
Bob then encounters an Articulated Robotic Arm, who gently lifts Bob and places it into a shipping box, working with precision to make sure Bob is all snuggled up to be sent to you. Finally, a Goods-to-Person (G2P) robot will retrieve the package from the packing area and deliver it to a human operator at the stationary pick station, who then performs a final check and adds necessary shipping labels. And now with the help of a delivery truck, you will meet Bob just in time for Christmas! Such technologies are common in warehouses owned by e-commerce giants such as Amazon.
The warehouse robotics market size is expected to grow considerably. In 2023, the market size is estimated to be US$6.74Bn, and estimated to balloon to US$15.22Bn in 2028. The fastest region is expected to be Asia Pacific, with major players such as Toshiba and FANUC.
Smaller players are also making a dent in the market. Exotec, a French start-up that specialises in G2P robots has recently raised US$335Mn in a Series D round led by Goldman Sachs. The company has now reached a valuation of $2Bn. The company produces a wide range of products along with robots, such as workstations, storage, monitoring software and smart conveyor systems.
With this expanding capability, we should see reductions in labour costs, streamlining in operations, increases in efficiency, optimisations in supply chains, and more Bobs meeting more lovely humans!
Predictive Analytics and Demand Forecasting
It’s no secret that since the turn of the century, the business world has become much more fast-paced. So, the question often comes down to: how do businesses stay ahead of the competition? Today, a part of that answer is Predictive Analytics.
In short, predictive analytics is a branch of data analytics that uses statistical techniques and machine learning to forecast future outcomes, allowing companies to see beyond the insights of the past and gauge possible future trends. This type of analytics plays a huge role in demand planning. In the process of forecasting demand, predictive analytics considers multiple variables and patterns to identify hidden demand patterns that we may not see, and uses them to predict future demand with a relatively high degree of accuracy. It achieves this by first analysing historical data to deduce a pattern, and then combine the data with information such as macroeconomic indicators, consumer behaviour, and competitor analysis to produce accurate forecasts.
Having access to such a wealth of information means that businesses are able to optimise different strategies. Firstly, predictive analytics will help businesses realise operational inefficiencies and bottlenecks, allowing them to correct any mistakes and optimise their supply chain and improve operational efficiency. Secondly, having full information of future demands is basically a dream come true for economists, who can then optimise pricing strategies. With price elasticity of demand at hand, economists can use this information to set prices that maximise revenue and optimise profit margins. And lastly, in continuation to the point above, economists can also realise different forms of price-discrimination, where predictive analytics assists in demand segmentation, thereby improving customer targeting and increasing profit-earning capabilities.
Now, one might observe that a potential drawback to this technology is the need to manage large sets of data. This might be very difficult without computer help. However, with the rise of AI and machine learning, we should see that in the future, this revolutionary tool should be able to allow businesses to shape a brighter future, allowing them to tap into profit-maximising strategies and economic principles that they previously did not have access to.
Natural Language Processing (NLP) for Customer Service:
Natural Language Processing is “a form of artificial intelligence that enables computer programs to process and analyse unstructured data, primarily free-form text data” . As a relatively new piece of technology, you can see why it’s used heavily in customer service. Most customer service requests are indeed ‘unstructured’: consider tweets, support tickets or emails. Thus, one has to deal with a great amount of messy data leaving employees to do a lot of manual and repetitive work. NLP has been the perfect tool to remedy this problem. It is often the case that NLP is used to create customer support chatbots, which, while being far from perfect, do allow for some two-way interaction between the company and customer.
What specifically makes customer service repetitive though? Simply we could look at it as a questions and answers exercise – the customer brings up a problem or wants to address a particular issue and the service rep tries to respond. Relying on their own intuition, the operator is overwhelmed by the pool of possible answers they could give and has to find the best one. Companies have so far managed this process by providing a closed list of problems and related answers, which the reps have to go through, sometimes manually. That is a slow and laborious process. NLP could provide that ‘best answer’ in an instant; it could also rank the answers based on their likelihood of resolving the issue. That means better response times and consequently a greater volume of resolved issues. While this approach can work quite well with the ‘list’ type of questions – i.e. relatively easy questions that can be answered with one precise answer – it may (and will) fail when confronted with complex questions requiring thorough research. Reps deal with these by analysing resolved past threads as knowing how a similar complex issue was successfully dealt with in the past is the best indicator they have available. NLP could automate this process too, by simply providing the best related historical threads. On top of this, NLP could group similar questions and thus provide mass answers to thus-created groups, appropriately route them to teams and people best suited to deal with them or prioritise high-value customers’ requests.
Some of the best startups working in this area are Yellow.ai. Founded in 2016, it has raised $102M by now, its most recent funding round being Series C It provides a “conversational AI platform for customer support”, “an omnichannel automation software to manage customer and employee experiences across channels” which aims at supporting companies with constructing their virtual assistants ultimately leading to an automation of operating workflows.
There is also PolyAI, a provider of AI-based voice assistants and thus also a customer-service-automating company, founded just a year later than Yellow and thus only had Series B as its latest funding round (and also only securing $70M);
outside the US there is Skit.ai – an Indian startup, which while also has gotten its Series B funding has only raised $31.1M by comparison. One can expect that as the international AI race picks up, we will see more and more such corporate creations in other countries, yet as of right now the market is pretty much concentrated in western countries and India.
Cybersecurity and AI in Logistics
“Cyber security has become an integral component of logistics infrastructure,” proclaims AAG on its website, “[as] digitization increases, so does the potential for cyber threats [and] the consequences of these threats are far from trivial – disruption and downtime can cripple organisations”. In their view, it is “emerging technologies like AI and blockchain” which will expand the possibilities of cyber security so as to detect risks in a quicker and more accurate way. As anyone familiar with ChatGPT knows, AI’s biggest skill is the ability to sift through enormous quantities of data and do with that knowledge what it may. In the cyber security context, this could mean identifying abnormal or malicious activities, such as zero-day attacks. It could also automate security processes, such as patch management, easing the process for many companies of fulfilling their security needs. Cybersecurity is essential in any logistics enterprise; after all, these are by definition supply chain companies and thus smooth functioning is their business model. The continuing integration of AI in cyber security solutions would mean greater microeconomic stability for a sector built upon that principle.
To move from theory to practice, let us look at the top 5 cybersecurity startups impacting logistics operating today. These are Gray Analytics, Moabi, ARX Alliance, CyberEQ and Cyberowl.
Looking at them in reverse order, Cyberowl is a Lonon-based startup selling cybersecurity analytics and compliance software to the shipping industry. The product is a cybersecurity monitoring and analytics system that enables cyber risk visualisation of onboard systems across the fleet. While that might seem like a lot of fancy technical mumbo-jumbo, it is a strong contributor to its (corporate) end-users’ cyber-security as it provides early warnings of cyber-attacks and monitors insecure behaviours, whether intentional or not.
On the other hand, CyberEQ is an Australian business intelligence solutions startup, mitigating supply chain risks for its customers. CISOSQUARED, its main product offering, anticipates insider threats within the supply chain itself by monitoring counterparties, 3rd parties, and other stakeholders. The risk management/analysis component seems to be a trend among these, as it is also the core business of Moabi and Gray Analytics.
Gray Analytics chooses a systematic approach. It first helps its customers identify the biggest potential threats and then provides the defensive systems needed to fence off the ecosystems at risk.
Moabi offers three product segments: a ‘supply chain’ component which means looking at the security of third parties involved in transactions (ie suppliers, sub-contractors, open-source); a ‘cybersecurity-by-design’ component meaning the implementation of security features so that bugs and weaknesses present do not trickle to clients and users; and, lastly, software audits.
The key takeaway here seems to be that the logistics is inherently based on stability; thus, cybersecurity is a necessity in the digital age for the businesses involved in the area meaning that the startups thriving in the space must engage in risk mitigation and analytics solutions as these are the goals of logistics companies anyway.
In the fast-paced landscape of modern commerce and logistics, the blending of warehouse robotics, predictive analytics, NLP in customer service, and AI in cybersecurity is reshaping supply chains. The warehouse robotics market, driven by AGVs and AMRs, is witnessing remarkable growth, while predictive analytics empowers businesses with foresight for strategic decision-making - preparing to bring the products you want to you, even before you knew you wanted them!
NLP, exemplified by startups like Yellow Freight and PolyAI, is transforming customer service by providing instant responses and efficient query categorization. Simultaneously, the integration of AI with cybersecurity, seen in startups like Cyberowl and Moabi, fortifies logistics infrastructure against potential threats.
These technologies collectively define the future of commerce, where efficiency, security, and customer satisfaction take centre stage. The true magic lies in their orchestrated synergy, ushering in an era of seamless and advanced goods delivery.