Swiss Partners
SERPICO Advanced Neural Technologies & Research Association, Suhr Switzerland
www.serpico.org
Philipp Jundt
Local Partner
SERPICO India Ltd, India
Sunterra, Sri Lanka
Project Type:
Technology:
Country:
Project Status:
Project Start:
End of Project:
Contract:
Swiss Contribution
The Swiss partner, SERPICO Advanced Neural Technologies & Research Association, is sharing AI and data systems with local partners to help farmers improve their access to credit, increase their productivity and enable the use of clean energy. SERPICO’s AI-driven credit assessment analyses farm and behavioural farmer data, such as satellite images and mobile records, to create fair credit profiles. This enables banks to lend to fish farmers with confidence, helping them to afford solar equipment. Swiss AI innovation is being put into practice on the ground, helping farmers to obtain loans, adopt solar power for their ponds and grow their businesses.
Description
India’s aquaculture sector, the second largest in the world, produces about USD 20 billion annually and supports 28 million workers, but many farms rely heavily on diesel generators due to unreliable or rationed electricity. Aerators are critical life-support systems for fish and must run 8–16 hours daily.
Diesel-powered aeration costs farmers USD 800–1,600 per pond annually, representing 20–40% of operating costs. Solar-powered aeration systems (4–6 kW with batteries) can eliminate fuel costs, provide reliable early-morning power, and deliver USD 6,000–19,000 in lifetime savings, with a payback period of 18–36 months. Despite these benefits, adoption remains below 3%, mainly because farmers cannot access credit, even though the potential financing market exceeds USD 5.8 billion.
Smallholder fish farmers are excluded from traditional lending systems because they lack formal financial records, credit bureau histories, and recognized collateral. Conventional credit models therefore classify them as unscorable or high risk.
Green ACRA (Alternative Credit Risk Assessment) addresses this gap by evaluating real operational behaviour instead of financial proxies. It measures factors such as production cycles, aerator usage, payment reliability to suppliers, satellite verification of pond activity, and community validation of farmers’ reliability.
The system converts these signals into a transparent credit score using a deterministic causal rule engine with more than 100,000 IF-THEN rules, developed from 8,800 surveys across 17 Indian states. Because the scoring is rule-based rather than black-box machine learning, every score can be fully explained and audited, allowing lenders to confidently present and justify decisions to credit committees.