By Dr. Maya Chen, Managing Partner · January 2025 · 11 min read

In March 2023, two researchers walked into our San Francisco office with a laptop full of soil sensor data and a conviction that the way the world farms is about to fundamentally change. Dr. Priya Nair had spent fifteen years as a soil scientist at the USDA Agricultural Research Service, developing some of the most sophisticated models for soil moisture prediction ever created. Her co-founder, Daniel Osei, had spent the prior decade building machine learning systems at Palantir for environments where data is sparse, latency is high, and the stakes are operational — not just analytical.

They were not pitching a startup. They were describing an inevitability. And we wrote the check forty-eight hours later.

The Problem AgriSense Solves

Smallholder and mid-scale farmers across North America and West Africa face a structural information deficit that costs them billions of dollars per year. They apply water, fertilizer, and pesticide based on calendar schedules, regional averages, and intuition — because the precision sensing and analytics infrastructure required to optimize at the field level has historically cost more than the incremental value it generates.

The result is systematic over-application. Farmers in California's Central Valley apply on average 40% more water than necessary for optimal yield. Farmers in Ghana and Senegal over-apply nitrogen fertilizer by 25–35%, not because they don't know better, but because they have no real-time feedback system to tell them when to stop. This isn't just economically wasteful — it is environmentally destructive. Agricultural runoff is the primary driver of freshwater eutrophication globally, and nitrogen overuse contributes to greenhouse gas emissions that are directly attributable to the farming sector.

AgriSense's solution is deceptively simple in concept and extraordinarily difficult in execution: deploy affordable IoT soil sensors across a field, combine their readings with satellite imagery and historical weather data, run the fused data through Dr. Nair's predictive soil models, and deliver precise, actionable recommendations to the farmer via SMS — accessible even in low-bandwidth, low-infrastructure environments.

Why We Invested: The Founding Team's Irreplaceable Edge

Many companies are trying to bring precision agriculture to smallholder farmers. What differentiated AgriSense from everything else we had seen was the depth and specificity of Dr. Nair's domain expertise and the sophistication of Daniel's approach to the data infrastructure problem.

Most precision agriculture platforms are built by technologists who approach the farm as a data collection problem. Dr. Nair built hers from the soil chemistry up. Her models for predicting optimal irrigation schedules account for soil texture, organic matter content, drainage characteristics, root architecture by crop variety, and microclimate factors — variables that simpler models ignore and that make the difference between a recommendation that's 70% accurate and one that's 94% accurate. That 24-percentage-point difference is the moat.

Daniel's insight was equally important: the delivery mechanism is as critical as the model. A recommendation system that requires a smartphone app, a reliable cellular connection, and a farmer who reads English isn't accessible to a smallholder in rural Senegal. AgriSense delivers recommendations via SMS in French, Wolof, Akan, and English — and the system is designed to work with $30 feature phones over 2G networks. This was not an afterthought; it was the core product decision that opened the entire African market.

"When Priya showed me the soil model predictions against actual field outcomes — 94% accuracy on irrigation timing, validated across three different soil types — I knew this was a different category of product from anything else we'd seen in agritech." — Dr. Maya Chen, Managing Partner, Sway for Future

The First 12 Months: Proving the Model

Sway for Future led the $5M seed round in March 2023. The founding team used the capital for three things: building out the hardware supply chain for their soil sensor modules, hiring four additional ML engineers with agricultural domain experience, and running a structured pilot with 500 farms in California's Central Valley — their proving ground.

The pilot results exceeded every projection. Average water consumption across the 500 farms dropped by 38% against the prior-season baseline — meaningfully higher than the 28% reduction the team had projected. Input cost savings averaged $840 per farm per year. And crucially, yields were unchanged or slightly improved in 92% of cases, validating that the reduction in inputs was genuinely optimized rather than simply undersupplied.

The Net Promoter Score among pilot farmers was 78 — extraordinary for an agricultural technology product and a reflection of the one decision the founding team made that we initially questioned: a dedicated "farm success manager" role for every 200 farms, responsible for onboarding, troubleshooting, and translating data insights into practical guidance. It was expensive at seed stage. It proved to be essential to the retention and expansion dynamics that now define AgriSense's growth model.

Expanding to West Africa: The Mission Becomes Global

The West Africa expansion was not in the original seed-stage plan. It happened because Dr. Nair had spent two years at USAID working on smallholder farmer programs in Ghana before joining the research service, and because the SMS-based delivery architecture Daniel had built made it technically viable from day one.

In January 2024, AgriSense launched pilots with 1,200 smallholder farms in three regions of Ghana and two regions of Senegal, in partnership with the Alliance of Bioversity International and CIAT (one of the CGIAR research institutions in our network that we introduced the team to after the seed close). The economics were different — average farm size smaller, sensor cost as a percentage of farm revenue higher — but the impact was more acute: for a smallholder farming two hectares in Ghana's Northern Region, a 35% reduction in fertilizer costs is the difference between a profitable growing season and debt.

By the end of 2024, AgriSense was operating on 12,000 farms in West Africa — 15% of the total network — with a fully localized product and a distribution partnership with Ghana's Ministry of Food and Agriculture that provides subsidized sensor access to farmers below a specific income threshold.

Post-Investment Value-Add: What Sway for Future Contributed

Capital is necessary but not sufficient. The three post-investment contributions we're most proud of at Sway for Future were introductions, not dollars.

Breakthrough Energy Ventures introduction (November 2023): We introduced the AgriSense team to our contact at Breakthrough Energy Ventures six months before their Series A process launched, allowing the team to build a relationship and demonstrate progress before any formal fundraising timeline. BEV ultimately led the $18M Series A in October 2024.

CGIAR Research Network connection: We facilitated an introduction to the leadership of CGIAR's Excellence in Agronomy initiative, which gave AgriSense access to fifteen years of field trial data across Sub-Saharan Africa — data that dramatically accelerated the calibration of Dr. Nair's soil models for African soil types and dramatically reduced the number of sensors needed to achieve prediction accuracy.

Bill & Melinda Gates Foundation engagement: Following the West Africa pilot results, we made an introduction to the Foundation's agricultural development team. In March 2025, the Foundation announced a $4.2M grant to subsidize AgriSense sensor deployment across 8,000 farms in Ghana and Tanzania as part of the Foundation's smallholder farmer income program — a non-dilutive capital source that extended AgriSense's runway while accelerating impact.

The Numbers Today

What Comes Next: 500,000 Farms by 2027

AgriSense's roadmap for the Series A capital is focused on three areas. First, hardware cost reduction: the team is now manufacturing its fifth-generation soil sensor module at a cost 62% below the original prototype, enabling economic deployment on farms as small as 0.5 hectares. Second, geographic expansion: Colombia, Nigeria, and India are the next three markets, each with existing distribution partnerships under negotiation. Third, the data product: the cumulative soil data from 80,000 farms is now large enough to begin selling anonymized regional soil health analytics to agricultural input companies, creating a high-margin revenue stream alongside the subscription business.

The 500,000-farm target for 2027 is ambitious. It implies a 6x growth from current scale in 36 months. But the hardware cost trajectory, the distribution infrastructure being built in new markets, and the Gates Foundation program creating subsidized entry points all make it credible. More importantly, it would represent — if achieved — more than $1.5 billion in cumulative annual farmer savings and a measurable contribution to food security across three continents.

That is the kind of investment we built Sway for Future to make.