Recapping the 2025 Regenerative Agriculture Science and Policy Forum, co-hosted by HDSI and Indigo Ag
“Agriculture and food systems are among the most urgent domains for advancing public-good outcomes through data science,” says Elizabeth Langdon-Gray, executive director of the Harvard Data Science Initiative (HDSI). This view shaped the June 2025 Regenerative Agriculture Science and Policy Forum, co-hosted by HDSI and Indigo Ag, a company that develops tools to support sustainable farming. The event brought together researchers, growers, technologists, and policymakers to explore how data science can accelerate regenerative agriculture and align environmental goals with economic incentives.
For HDSI, co-hosting this forum extended a growing body of work connecting data science to agricultural resilience, including projects on climate-linked malnutrition in Madagascar, food supply volatility, and a USDA-partnered datathon where students addressed challenges in real-world agricultural data. These efforts reflect HDSI’s broader commitment to applied, interdisciplinary impact. Agriculture presents many of the same core challenges seen in other domains—integrating fragmented datasets, building transferable models, and addressing gaps in measurement and interoperability. The complexity of these problems demands a multi-sector approach; coming together with Indigo Ag offered a unique chance to connect methodological innovation to real-world agricultural impact.
Over two days of discussion and collaboration, participants identified technical hurdles and actionable strategies for scaling regenerative practices amid climate change. Below, we share some common themes heard throughout the Forum’s sessions.
Aligning Stakeholders Across the Food System
Efforts to scale regenerative agriculture often struggle due to misaligned incentives and fragmented communication among those involved—from researchers and funders to farmers and regulators. These gaps are deepened by differences in how each group defines evidence, timelines, and success. For example, farmers need reliable outcomes within a growing season, whereas researchers may focus on long-term statistical trends. Participants underscored the need for shared language and intentional spaces for sustained dialogue, especially to build trust with growers.
Opportunities:
- Creating standards for how outcomes like soil health, resilience, and productivity are measured and reported.
- Enable the design of multi-layered platforms that translate technical data into tailored insights for different stakeholders—scientists, farmers, investors, and regulators—promoting a more integrated understanding of impact.
Climate Resilience and Yield Stability
Extreme weather events—particularly heat waves—are already driving down yields, making climate resilience a top priority. But identifying which practices reliably build resilience is difficult, as regional variation and inconsistent datasets obscure trends. Farmers often rely on local experimentation, which remains disconnected from broader research or policy. Forum participants emphasized the importance of farmer-generated data and experience, alongside technologies like remote sensing, to fill these gaps.
Opportunities:
- Expanding use of satellite imagery and time-series data can enhance resilience efforts by detecting early stress signals in crops and associate them with specific farm practices.
- Ensemble models can integrate multiple predictive techniques to reflect local variability, helping to inform both adaptive agronomic strategies and policy mechanisms such as insurance reform.
Accounting for Greenhouse Gas Emissions
Efforts to quantify emissions and carbon storage in agriculture must account for scientific and methodological uncertainties. Programs to account for the emissions impacts of regenerative agriculture must address key questions: How should carbon be attributed in rotating crop systems? How are reversals—when carbon is lost—tracked on shared land? Inconsistencies in addressing these questions across methodologies and frameworks can cause confusion for growers and companies trying to address their environmental footprint.
Opportunities:
- The development of tiered accounting frameworks, suited to different confidence levels, can provide structure and transparency to this evolving space.
- Shared calibration datasets, standardized reporting templates, and peer-reviewed models with documented assumptions can support both credibility and flexibility in land-sector GHG inventories.
Rethinking Permanence and Carbon Contracts
The concept of permanence—storing carbon for 100 years or more—is challenging to assure within the realities of agriculture, where land use is shaped by shifting markets, weather events, and generational decisions. Farmers are challenged to reasonably commit to long-term carbon storage without financial or operational risk. Forum participants advocated for alternatives such as “ton-year” crediting, which values shorter-term sequestration, and flexible contracts that reward stewardship without rigid timelines. Managing permanence as an aggregate across a large population provides one route to ensure project permanence while providing farmers with flexibility.
Opportunities:
- We can enable these alternatives by modeling time-bounded carbon storage and crediting approaches that reflect observed practice changes.
- Algorithms that flag early signs of reversal or non-compliance using remote sensing and management data can support conservative but realistic credit accounting, building trust without overpromising.
Moving Beyond the Basics: Emerging Soil Innovations
While practices like cover cropping and no-till have gained attention, newer approaches—biochar, enhanced weathering, beneficial fungal seed treatments—could add more impact per acre with agronomic and environmental benefits. These technologies often lack integrated verification protocols and face barriers such as high initial costs and unclear market pathways. Participants highlighted the potential to stack multiple practices for greater effect but noted that current systems aren’t designed to capture co-benefits.
Opportunities:
- Technical scaffolding for more complex interventions by developing modeling frameworks that track and quantify the joint impacts of multiple soil innovations.
- These models can inform verification protocols that credit synergistic benefits while preserving conservative baselines, unlocking new financing pathways.
Policy Realignment and Funding Pressures
Policy environments present a mix of obstacles and opportunities. At the federal level, conservation programs are facing budget cuts that reduce technical support for farmers. Innovative state-level programs are emerging, but their design and focus vary. Political hesitancy around climate framing further complicates the picture. Participants argued that positioning regenerative practices in terms of water quality, yield stability, or rural economic benefits can help sustain bipartisan support.
Opportunities:
- Causal modeling of different policy framings—quantifying, for instance, how practices affect water use or profitability—allowing advocates to tailor narratives to local and political contexts.
- Shared metadata and interoperable formats also enable coordination across programs and jurisdictions, increasing overall system efficiency.
Turning Data Into Action
Despite advances in soil measurement tools, data collection remains costly and time-intensive, especially for farmers. Even when collected, data is often siloed across incompatible systems. The underlying issue is that tools are typically built either for researchers or regulators—not the end-users managing the land. Participants expressed a strong desire for systems where growers only need to enter data once, with outputs usable across agronomic, financial, and environmental contexts.
Opportunities:
- Developing interoperable platforms and application programming interfaces (APIs) that allow single-entry, multi-use data flows.
- Harmonized metadata and model-layer translation can ensure that data collected for one purpose (e.g., practice tracking) can also inform others (e.g., crediting, extension support, or financial planning).
Connecting Food Security and Computing Infrastructure
Regions facing the greatest food insecurity—particularly in parts of Africa—are also those with the least available data to guide agricultural policy. Sparse sensing infrastructure and low data availability make planning difficult. But waiting for perfect data is not an option. Participants stressed the importance of building local partnerships and supporting adaptable models that can work with limited inputs.
Opportunities:
- High-performance computing and federated data science approaches can fill this gap by enabling predictive modeling even under conditions of sparse or noisy data.
- Federated learning allows models to improve over time while maintaining local data privacy, offering a path forward for equitable and scalable food system resilience planning.
Closing Thoughts
There is growing alignment around the importance of data-informed, farmer-centered solutions, but much remains to be done to convert insights into reliable practice. Data science offers a critical foundation for transparency, adaptation, and trust. With persistent effort and shared ownership, the path toward scalable, equitable regenerative agriculture is starting to come into view.