Our Investment in Vivid Machines: Building the World’s Largest Plant-Level Dataset in Specialty Crops
StandUp Ventures is excited to announce our latest investment in Vivid Machines’ oversubscribed $4.3M USD Seed round. We are partnering alongside a fantastic group of investors including BDC Capital’s Thrive Venture Fund, TallGrass Ventures, Connexus Venture Capital (Emmertech), the W Fund, Cornell University, Freycinet Ventures, BoxOne Ventures, EntrepreneurFirst, MaRS IAF, and the owner one of Canada’s largest apple producers, Kirk Kemp from Algoma Orchards.
As venture investors, the recent remarkable progress in AI has led us to think deeply about the types of companies that possess a strategic, compounding, and defensible edge in the rapidly evolving “new normal.”
Lately, we have been particularly interested in companies that solve optimization problems where the collection and utilization of data has historically been difficult. Our thesis is that if data collection hurdles can be overcome, the application of machine learning in these previously under-digitized spaces often has a clear capacity to drive immediate and immense value.
With developments in passive data collection techniques and AI model development, it seems as though we are at a precipice when it comes to driving value with technology that can solve optimization problems in practical business contexts.
We’re excited to share how our latest investment fits this thesis in the world of fruit production!
Optimization in Nature
One example of a natural and dynamic optimization problem is the growth of a tree, and specifically — a fruit tree.
With finite energy, a tree allocates wisely (a new branch, a new blossom, a bigger blossom perhaps?) — dynamically trading off energy waste during the process of growth, with some probability that the seeds will land in fresh new soil.
Because of this naturally optimized design, employing interventional techniques during the growing season can make a large difference in terms of the amount and quality of fruit that is produced.
To professional fruit growers, this highly technical optimization problem is called crop load management. If this can be solved, there is an enormous opportunity to drive profitability and real value across the fruit production supply chain.
What is Crop Load Management?
To enhance quality and size of fruit, growers often prune branches and buds, and thin blossoms and green fruit during the growing season to limit the number of fruits on a tree. This reduces competition for energy between fruits, provides ample space, and allows sufficient sunlight.
Inadequate thinning can lead to lower quality fruit and more resources required later in the season, while removing too many fruit can result in lower yields and storage issues. And if that isn’t enough to consider, a successful harvest sometimes puts too much burden on a plant, reducing capacity for the crop that follows. Finding the right balance for all of these factors is critically important for maximizing grower profitability.
Today this is difficult, largely due to the availability of quality data that can drive these decisions. The status quo for data collection involves manually counting the number of buds or fruitlets in a small sample of an orchard. These population samples are time-intensive to acquire, expensive, not always representative of a crop broadly, and prone to human-error. It’s also near-impossible to collect and action the data in real-time.
Because of this, many growers, and especially the smaller ones, don’t utilize crop load management techniques at all.
And even with improved data collection, the dynamic and multi-variate nature of growth makes it challenging even for the growers with the best information to make the right decisions. Some growers do use outdated and naïve models developed by universities, but the advancement of these has been difficult without good foundational datasets.
All of this inefficiency and it’s no surprise that grower profitability has become a burning pain-point, with 50% of fruit farmers having lost money every year since 2013.
The Solution
Founded by Jenny Lemieux and Jonathan Binas, Vivid Machines’ vision is to build the world’s largest per-plant dataset in specialty crops using proprietary hardware and AI-powered software that captures fruit-level data in real-time during the growing season. This data, coupled with intelligent analysis, empowers farmers to make informed decisions to improve their crop yield.
By providing real-time, actionable, fruit-level data, the company’s proprietary hardware and AI-driven software offer a range of products that can support growers. These include fruit quantity and size estimation, growth rate prediction, and in the future will provide early pest and disease identification, as well as nutrient deficiency monitoring, all of which significantly expand the potential market opportunity.
And growers aren’t the only folks in the supply chain who stand to gain from data-driven optimization. Packhouses and sales desks require early, accurate predictions from growers to manage costs and avoid losing contracts. Supply and demand are rarely aligned, and marketers that negotiate grower and packhouse contracts with large distributors don’t often have access to numbers until mid-late fall after apples have been harvested and packed. As a result, they often default on contracts because they can’t deliver on what is promised. Serving these stakeholder pain points further increases potential revenue streams for the company.
The Opportunity
The market is massive — with the global fruit and vegetable industry being valued at over $1 trillion and rapidly growing. Although Vivid Machines is starting with a beachhead market of apples, the company is quickly expanding its offerings to cover the top 15 tender fruits, representing an opportunity of over $15 billion.
While some competitors have tried to use computer vision for crop load management, their lack of real-time data that sets the foundation for better model development has so far limited the opportunity for optimization insights in the space. Because of this, Vivid Machines has gained impressive traction in this whitespace after initial pilots demonstrated efficacy rates of over 90%.
“By the end of the season last year, yield prediction accuracy was on average 90%, which gave us the confidence to commit to working with Vivid as we know their technology can help us improve yield, and quality, increasing sales,”
- Kirk Kemp, owner of Algoma Orchards
Through the application of cutting-edge machine learning to their proprietary data, Vivid Machines can generate previously unattainable insights that drive efficiencies for growers, providing net new value to stakeholders throughout the industry.
Jenny and Jonathan bring a wealth of diverse experience and expertise to the table. Jenny holds degrees in chemical engineering, design, and data science, and has over 15 years of experience building physical and software products for startups and large corporations. Meanwhile, Jonathan brings a strong background in physics, brain-inspired hardware, deep learning, and accomplishments in agri-tech research.
We have been so impressed by Jenny and Jonathan’s massive ambition, powerful execution, and stellar reputation with key folks in what is typically a relationship-driven industry. We have very strong conviction that their combination of technical acumen and business talent is well-positioned to execute in the food data space and build a large and sustainable company that drives real value for the industry.
Warm welcome to the portfolio Vivid Machines!
