SWARM Engineering: Driving The Democratization Of AI In The Agri-Food Supply Chain

Artificial Intelligence (AI) has become a part of our lives. This disruptive technology has transformed various business operations and automated industries. Although it is still considered an unattainable goal to some businesses, the technology has made its journey towards success. But how accessible is it for business users? Not enough. SWARM Engineering has a solution.

SWARM is transforming the way people solve problems, with a vision to democratize the use of AI and make it easily accessible to everyday business users. The company is focused on tackling inefficiencies in the agri-food supply chain, reducing food waste, and lowering carbon footprints, by finding more effective ways to operate processes such as forecasting yield or balancing supply and demand. SWARM is a Software-as-a-Service (SaaS) platform designed to automatically solve challenges with a curated library of the latest algorithms. Its services help customers easily define any challenge in a way that the SWARM software can solve automatically.

A Business Born Out of Sheer Imagination

The original concept of SWARM came from a conversation between SWARM's CEO Anthony Howcroft, and Dave Bartlett, who at that time was the CTO for GE's Power and Light business (he is now a member of SWARM's Advisory Board). The discussion imagined what Facebook for machines might look like, and how machines could swap information to solve problems. For instance, would a jet engine ask other engines about the occurrence of intermittent vibration at a certain temperature and humidity? Perhaps the engines would discuss where they had traveled, what type of aircraft they were attached to, recent weather conditions encountered, ultimately identifying causal factors of the issue together.

This initial concept sparked an investigation into state-of-the-art multi-agent systems and AI. The company realized that by combining these technologies with the waves of data beginning to stream off IoT devices, there was a great opportunity to produce a system that could learn to solve challenging problems for organizations. Having come from a background in data warehousing, the SWARM team understood the value of analytics in identifying actionable insights. The team also suspected there was another layer of capability, where problems could be solved rather than highlighted for users to fix. For example, instead of pointing out a fungal infection and treatment required for a damaged crop, it would be better to determine in advance the combination of plant density, humidity, temperature, and CO₂ that are conducive to fungal infections and optimizing the environment to keep it safe from such issues. SWARM's goal from the start was to move higher in the analytic maturity model and solve problems rather than offer remediation.

The first pilot project involved a major food producer that invited Swarm to solve product blending issues. The customer wished to mix a handful of elements from 200 ingredients that were seemingly identical but differed in quality, and it turned out there were over 85 trillion possible combinations. Every day the company took a blending decision and the choice had a significant impact on cost and customer experience. The original spreadsheet system was being overwhelmed, and SWARM stepped in to solve the problem. It did this by running a multi-agent mating algorithm, which could make the daily blend recommendation in a couple of minutes while minimizing cost and meeting both taste and FDA health requirements.

SWARM delved into similar scenarios and discovered that supply chains were filled with such challenges, and the agri-food supply chain had some special characteristics that made it ripe for innovation. It was clear to SWARM that the Agri-Food sector was and still is undergoing a huge digital transformation, with new sensor technology becoming omnipresent in every aspect of the value chain; from monitoring crop and greenhouse conditions to tracking logistics. Simultaneously, consumer demand is forcing new behavior, such as alternative proteins, on-demand shopping and delivery, traceability, and the desire for local and sustainable produce. SWARM notes that established firms are being forced to transform and adapt to compete with a wave of new, technology-enabled entrants that are themselves experiencing tremendous growth.

A team with Microsoft pedigree

Anthony Howcroft, CEO and founder of SWARM began his career as a software developer at Kraft Foods, working on production planning systems for coffee beans, and has subsequently worked at a mix of startups and big corporates. His formal education was at the University of Oxford, where Anthony studied Creative Writing and went on to become a prize-winning author of fiction, as a hobby alongside his technology career. Last year, he published his first non-fiction book 'Questions – A User's Guide', which became a top-ten Amazon bestseller in its category. The book took seven years of research, and many elements are being incorporated into the SWARM software, especially in the way the company captures and models problems. Einstein once said, "The formulation of the problem is often more essential than its solution", and SWARM has dedicated significant effort to making software that lets business users easily capture definitions of their problems – or challenges, as SWARM prefers to call them. Anthony says, "Nobody wants to list the things that are wrong with their business, the problems. We help find areas we can improve, challenge the business to do more, to be better."

All three C-Level executives at SWARM come from a Microsoft background. The company's Chief Data Scientist, Shiyi Pickrell, worked at Microsoft for 8 years as Director of Data Science and Machine Learning, and at Amazon before that. SWARM's CEO and his Chief Revenue Officer Andy Mouacdie,  were both at DATAllegro, the data warehouse appliance vendor acquired by Microsoft, and were a core part of the team leading Microsoft's global big data initiative. It is no surprise to see that SWARM has recently announced their availability on the Microsoft Azure marketplace, and are in the process of gaining certification for key Microsoft partner programs.

AI and Big Data as the Key Drivers

SWARM is driving the democratization of AI in the food supply chain. The company reveals that the benefits are typically measured via an ROI of at least 4x, which translates into multi-million-dollar savings. Often, though,  the biggest saving is time. Many of the processes SWARM optimizes are those that regularly take hours of effort from teams of domain experts and those in operational roles, like planners. With so much data to analyze, often from disparate sources, the company believes that it can be a time-consuming and onerous task to make decisions. SWARM's software can do the heavy lifting, resulting not just in better plans, but hours of time saved for key personnel in their customers, which can be dedicated to other high-value initiatives.

 Strong Partnerships behind the Success

According to SWARM, innovation is a way of thinking, an attitude, and relies on speed.  Startups can often out-turn their competitors because they make and implement decisions faster than bigger firms. It also helps that a startup team is fully committed because everyone has a stake in the company's success. "Being in a startup is really a lifestyle choice. For the right person it is a huge opportunity, and a rewarding environment," Anthony says. "We are lucky to have a fabulous team who are fully committed to our customers' success."

SWARM has three critical partnerships: first, the customers who have been willing to share their challenges and have been open to making a change. Second, are the academics who have guided the company on the latest machine learning and algorithm research. The Advisory Board comprises experts like Naira Hovakimyan, a world-class scientist who won the Humboldt prize for lifetime achievements, Chris Watkins, who invented the Q-learning algorithm that kick-started the resurgence in deep learning, and Lawrence Henesey, who is a professor in both AI and maritime logistics. The third critical group is the investors who are vital partners, with venture firms such as S2G Ventures and Serra Ventures, along with key industry luminaries like John Power from the Harvard Business School Alumni, with deep connections into the agri-food supply chain, who have been instrumental in making the right introductions in a market where word-of-mouth is so critical, the company states.

Big Pockets Capitalizing on Talents

Early on, SWARM constructed a tool that allowed data scientists to rapidly build and deploy innovative solutions to classic supply chain problems, using advanced AI and machine learning. As the company validated its approach with potential clients, SWARM discovered a scarcity of data scientists within its target customers. SWARM discloses that only the largest firms had such talent at their disposal, and they were frequently a bottleneck. Instead of pivoting as such, the company created new tools built on their core platform, specifically for business users, to both define their challenges, and execute the solutions, using the original SWARM Solution Engine.

From Disbelief to Recognition

Some of the initial companies that SWARM spoke to were concerned that they did not have enough data on key aspects of the business, or that their data was not clean enough. The company states that although data quality can always be improved, and data integration is always time-consuming, the real challenge is often one of perception. Here, people find it hard to believe that a process that has been in their business and continually evolving for twenty years or more can be improved by 10-20%. The initial reaction to many of SWARM's proof-of-concept projects has been disbelief, and often the solutions are scrutinized and heavily tested before clients can accept their veracity, at which point they move on to delight and the focus rapidly switches to implementation.

SWARM reveals that it had one C-level executive at a major US port who declared that the funds used on the company's pilot project were the best money he has ever spent, after thirty years in the industry. SWARM has started gathering some industry awards recently and was recently named one of the Top Ten Cognitive Solutions to Watch in 2021. Although, according to the company, the best reward is seeing the value they can bring to a customer, and gaining follow-on projects. One of SWARM's earliest customers now has four projects underway to improve multiple aspects of its supply chain.

Technology will Yield Quality and Sustainability in Future

SWARM commented that the future will witness more disruption, more black swans, and there will be data and AI wars, where businesses rise and fall based on the quality of their algorithms. The company brings up the scenario when Google's AlphaGo AI algorithm beat Lee Seedol, the 18-times World Champion at Go, and there was a huge press fanfare. A year later, a new algorithm called AlphaGo Zero beat the previous algorithm by a score of 100-games to zero after learning to play Go in 3 days. SWARM conveys that if a company had embedded the original algorithm in its supply chain software, and a competitor had waited a year and used the new algorithm, then the first company would be in a tough spot. This algorithm war will be the new normal.

As SWARM observes, organizations that are buying applications with embedded AI may believe they are doing the right thing, but they could be signing a death sentence. The company reiterates that firms need to stay flexible, focus on the challenges they want to meet, and separate that from the solution. SWARM believes that it provides a great platform to achieve this, which is delivering results now, without taking a risky bet on the future. By separating the challenge from the solution, SWARM can test and switch in new algorithms when required, keeping an organization at the cutting edge of the algorithm war. On a concluding note, SWARM mentions that while ultimately some companies will suffer, the result for consumers will be positive. The company believes that people will see higher quality, healthier food, delivered at a lower cost in a more sustainable fashion. The future is bright, as AI and humans collaborate to improve the agri-food supply chain.