2026-07-11 –, The Seedbed
As proprietary technology systems increasingly deploy artificial intelligence to centralize agricultural advice and enclose farm data, how can open-source networks utilize these same tools for community empowerment? This 60-minute session moves away from high-level theoretical debates to look at real-world case studies of AI applied to agroecology. Opened by Anamika Dey, the session features four short presentations illustrating how open-source software, voice-first data processing, and localized models can protect indigenous knowledge, automate documentation, and elevate the lived expertise of land stewards.
This consolidated programming block features direct project insights from four international fellows deploying localized AI frameworks. The session moves intentionally from an alternative baseline framing into four distinct operational case studies. It details the technical development of context-filtered "curiosity engines," voice-first Standard Operating Procedure (SOP) logs for organic certification compliance, offline-first translation arrays parsing local dialects, and decentralized machine learning pipelines designed to prevent data harvesting while securing knowledge provenance.
Session Rundown / Outline:
- 00–10 mins: Opening Framing (Anamika Dey) — Setting the stage by contrasting top-down "virtual agronomists" that push rigid chemical recommendations with decentralized, community-governed AI tools designed to prompt localized experimentation. Outlining how open networks can repurpose machine learning architectures for smallholder data autonomy.
- 10–20 mins: Case Study 1: The GIAN Curiosity Engine (Anamika Dey). A technical deep dive into the GIAN and Honey Bee Network's prototype of an AI-powered annotated database system. Drawing on work developed with Kushal, Anamika will share how they connected 6–7 distinct datasets (including the UNDP database) and deployed a contextual agroecological filter. Instead of providing automated flat answers, this "curiosity engine" generates heuristic questions to spark local trials, introduces interactive trial cards after successive searches, and maintains strict source attribution to protect grassroots inventors.
- 20–30 mins: Case Study 2: Voice-First SOP Capture & Compliance (Kirsten): Showcasing recent field implementations from Farm Flow. Kirsten will demonstrate how they are utilizing lightweight voice-to-text processing and OCR tools to solve ag-tech's biggest barrier: getting accurate field data into a system without disrupting a farmer's workflow. The talk outlines how farmers use voice notes to instantly document Standard Operating Procedures (SOPs) and maintain auditable paper trails for organic certification, showing how algorithms can support and preserve a farmer's judgment rather than trying to replace it.
- 30–40 mins: Case Study 3: Protecting Bioregional Knowledge (Larissa): Unpacking the practical applications of locally hosted, offline-first large language models (LLMs). This presentation details how small-scale networks are translating multimedia data inputs and local dialects to make expert agtech lingo accessible across boundary lines without extracting sensitive community data to commercial clouds.
- 40–50 mins: Case Study 4: Decentralized Intelligence Architectures (Benchadid): Exploring how decentralized machine learning pipelines can aggregate unstructured field observations and traditional ecological knowledge. The presentation focuses on building secure data tokens and technical attribution layers to ensure that value cycles back to the original knowledge holders.
- 50–60 mins: Open Discussion + Q&A
Dr Anamika Dey is the CEO of GIAN (Gujarat Grassroots Innovation Augmentation Network) and a researcher working at the intersection of traditional knowledge sovereignty, community data governance, and grassroots innovation in India.
For over a decade, she has worked with the Honey Bee Network — one of the earliest models of community-owned knowledge documentation, built on principles of recognition, reciprocity, and benefit-sharing with knowledge holders. Long before "data commons" became a tech conversation, the Network was asking: who owns what a community knows, and who profits from it?
She brings this lens to questions of Community Data Sovereignty, ethical bio-entrepreneurship, and what decentralised governance of knowledge actually looks like on the ground — in villages, not whitepapers. She is also the founder of The Little Himalayan Co., a social enterprise that attempts to close the loop between knowledge holders and markets without extracting value from either.
She is interested in finding how decentralised architectures can serve communities that have the most to lose from centralised platforms — and the most to contribute to a more equitable web.
Co-founder and technology strategist with Farm Flow. She focuses on human-centered UX design, voice-first data logging ecosystems, and open-source organic compliance software.
Electronic and Computer Engineer from the Federal University of Rio de Janeiro (UFRJ). Master’s student in the Graduate Program in Technology for Social Development at the Interdisciplinary Center for Social Development (NIDES/UFRJ). Extension researcher at the Technical Solidarity Center (SOLTEC/UFRJ). Software developer at the EITA cooperative, developing free/libre software technologies for grassroots social movements, and collaborating developer at TEKOPORÃ, a free software collective dedicated to creating community-oriented management solutions guided by agroecological principles.
Agricultural technologist and founder of Green Leaf AI, specializing in decentralized data provenance and deploying community-governed data applications under low-connectivity constraints.