NEWS: AI Software Company Seeking $2 Million. And, It’s Been Chosen to Participate in a Highly-Selective Global Food and Agriculture Accelerator
A company designed to solve complex optimization challenges using industrial sensor technology is seeking $2 million in bridge funding before opening a Series A round.
The startup has also recently been selected by Terra, a global food and agriculture accelerator, which officially launches its third cohort next week at the San Francisco location of RocketSpace, a co-working company that created the accelerator.
Swarm Engineering, with HQ in Irvine, has already raised $1 million in seed funding from investors, including the Harvard Business School Alumni Angels.
(The name of the company is based on an allusion to complex problems in nature being solved by cooperative behavior in a swarm of insects.)
How it Works
Swarm’s tech represents the next generation of cognitive computing, combining a multi-agent system with what’s known as “reinforcement learning,” a subset of machine learning. Similar to how children are taught, reinforcement learning involves both positive and negative reinforcement.
Swarm uses reinforcement learning algorithms to optimize supply chains for sectors including food production and logistics.
Swarm has obtained one global food customer, which it cannot disclose, due to confidentiality agreements, and is in the midst of a pilot project with a second customer, GrainCorp with HQ in Sydney, Australia.
For a complex process like blending in food production, Swarm created a software agent that uses a “mating” algorithm that yields a “very optimal” scenario in real time, even when the process is “highly disrupted and dynamic,” CEO and founder Anthony Howcroft told OC Startups Now.
For example, if a chocolate company wants to create a new bar that’s 80% dark with chocolate from Ecuador, but its vats are filled with cocoa beans from all over the world, it needs to do this in the most optimal, cost-efficient way. Ecuadorean beans may be the most expensive so the least amount of these beans needs to be used, while keeping the distinctive flavor of that region.
Mixing in too many beans from other locations, or with different characteristics, may ruin the blend. An Excel spreadsheet can’t handle the number of permutations possible – trillions of options. Swarm’s technology can, Howcroft said.
Swarm’s two products are NestView, a modeling and simulation tool which investigates and defines the blending or logistical challenge, and AgentFlock, a multiplicity of small software agents that are then deployed to solve the challenge.
The company is looking to get $2 million in bridge funding to hire more engineers and support its existing customers. It also wants the funds to boost its valuation for a planned Series A raise next year, by getting more customers from a “rapidly growing” pipeline of opportunities.
Swarm currently has four full-time employees who work out of the WeWork co-working space at 200 Spectrum Drive in Irvine.
Competition in blending optimization includes AspenTech and Honeywell.
Swarm is more of a general platform for “multiagent systems,” so it competes more directly with other AI startups focused on agent technology, Howcroft has said.
Partners and Advisors
The company is partnering with two other firms: ThinkIQ in Aliso Viejo and Maana, which has offices in Palo Alto, Houston, and Bellevue, Wash. Both have strong data platforms and clients looking to optimize blending solutions, Howcroft has said.
Swarm has recruited an advisory board that includes Dave Bartlett, CTO at Panasonic Avionics, and Lawrence Henesey, professor of maritime logistics in Sweden, with more than 27 years experience of using multi-agent systems in port environments.
Its board also includes Naira Hovakimyan, the “world-leading” control-theorist, awarded the Humboldt prize for lifetime achievements, who has advised NASA and the Dept. of Defense on automated flight control. And, another board member is Chris Watkins, a professor of AI at Royal Holloway University of London, and inventor of the “Q-Learning Reinforcement Learning algorithm” that has driven much of the recent interest in reinforcement learning as a machine learning technique for the Internet of Things (IoT) age. The IoT is is the network of physical devices, vehicles, home appliances and other items embedded with sensors and connectivity capabilities, which enables them to connect and exchange data.
Swarm was chosen to participate in Terra’s accelerator amidst stiff competition, being selected as one of the 16 “most innovative startups” from across the globe.
Terra’s accelerator model is designed for later-stage startups that are looking to validate and grow their business with industry-leading corporations. And, it doesn’t take any equity in these startups.
At the accelerator, Swarm will focus on optimizing critical aspects of GrainCorp’s supply chain over the next three to six months, as part of a broader innovation program for the international agribusiness. GrainCorp’s diversified operations span four continents and supply customers in more than 30 countries.
“We are excited by the potential in Swarm Engineering to continue the evolution of our supply chain process, to deliver additional value for our farmers and buyers, and in helping our employees manage complex scenarios,” GrainCorp CDO Marcus Kennedy said in a press release.