Introduction

Dymaxion Labs stands at the intersection of AI and geospatial analytics, driving change with data-driven solutions. Our platform, fueled by AI, reshapes decision-making by providing real-time insights for informed policy. Open Source libraries amplify the potential of machine learning models, effortlessly processing satellite and drone imagery on the cloud.

The agriculture and food market in Latin American and Caribbean countries holds significant global importance. We monitor vast terrains, predicting challenges like floods and droughts, safeguarding food production. The urgency is real as we address health and nutrition disparities in these regions—with chronic undernutrition severely hindering children's future prospects—aiming for a more equitable future.

Our platform empowers companies, governments, and NGOs to access up-to-date, readily available data, facilitating the measurement of public policy impact on food security and household conditions. By leveraging this data, decision-makers can make informed choices that positively influence their populations and shape their futures. Use cases go beyond agriculture and can be parlayed to urban areas as we map huge swathes of informal zones. The potential is tangible: real-time weather monitoring, disaster preparedness, and urban planning fueled by innovation, not guesswork.

In 2021, we received bridge funding from the Venture Fund to unlock the potential of satellite imagery and promote geospatial and data-driven approaches worldwide. The funding aimed to conduct pilots focused on food security, poverty, and climate action, exploring the synergies between earth observation and blockchain technologies.

Milestones

During the course of the Venture Fund’s growth investment in the past 12 months, our team was able to evolve our Open Source libraries, allowing customers to run their machine-learning models for satellite imagery processing without worrying about geospatial libraries installation and infrastructure scale. As part of the project, this tool, Dymaxion Labs Toolkit (DLT), became a Digital Public Good, which was an important milestone for our product development roadmap. We mapped 1,750,000 square kilometers across Argentina, Brazil, Colombia, Honduras, Perú, and the US.

Additionally, the tool gathers a set of libraries developed in the last years with the goal of accelerating the training and inference of machine learning models for geospatial imagery in agriculture and urban development domains. The set of libraries consist of the following: satproc for satellite imagery processing, unetseg for training and inference with the UNET deep learning architecture, and labfunctions for deployment and scaling.

An organization can therefore deploy new infrastructure using labfunctions, process its own data there with satproc to get it ready to be ingested by a neural network, and finally apply the model at a massive scale in their area of interest.

Collaborations & Partnerships

To enhance our solution, we collaborated with the following three organizations over the past 12 months to expedite their ground surveys and gather feedback that would inform the next steps in our product roadmap:

• Coopecan (AgTech)
  We worked closely with Coopecan, an AgTech organization, to develop and implement a machine learning-derived biomass availability measurement. This initiative aims to support sustainable agricultural practices in livestock farming within the Peruvian Andes region. The algorithm's outcomes are securely stored on a public Blockchain for transparency and accessibility.
   
• TECHO in Brazil and Chile
  In partnership with the NGO TECHO, we trained their teams in both Brazil and Chile to create an algorithm capable of monitoring informal settlement growth in peri-urban areas. The algorithm leverages public satellite imagery to track and analyze the expansion of these settlements, aiding urban planning efforts and social development initiatives.
   
• AI Climate Working alongside AI Climate, we successfully trained a machine learning algorithm to assess and map informal settlements' vulnerability to flooding and landslides in the Sula Valley of Honduras. These algorithms are being validated by local communities and are intended to be integrated into a climate change analysis platform developed by the I2UD, thus empowering better disaster preparedness and mitigation strategies.

These collaborations have been instrumental in advancing our product and have paved the way for meaningful contributions in agricultural sustainability, urban development, and climate resilience.

In parallel with these collaborations, we had been working on amplifying the impact of our tool with a series of webinars, virtual workshops, and training sessions oriented towards developers, urbanists, and data scientists looking to add value to their projects. Recordings and relevant information can be found in our blog, available in Spanish and English.

We invite you to collaborate with us by sending us an email to contact@dymaxionlabs.com and contributing with our open-source tools in GitHub.

Acquisition

To conclude these exciting 12 months, full of learning and experimentation, we announced the acquisition of Dymaxion Labs as part of GDM Seeds (GDM). Dymaxion Labsʼ founders believe that this movement is a massive step forward for the company, in order to continue developing tools that can help organizations around the world to meet the SDGs. As part of the agreement, Dymaxion Labs will continue operating independently, focusing on the verticals of sustainability practices in agriculture, food security, and climate change.