CK Prahalad

Agricultural intermediaries still bundle credit, inputs, and marketing into exploitive dependency traps. Farmers cannot refuse any single element without losing access to all others. Digital agricultural platforms have proliferated, but they address pieces—price information here, credit there—without competing against the trader's bundled offer.

An AI-orchestrated platform that provides the complete bundle—credit scoring based on satellite imagery and soil data, input procurement with quality verification, real-time price discovery across multiple markets, and guaranteed offtake—delivered through local sanchalaks (village operators) who take public oaths and earn commissions. The system must be self-sustaining through procurement margins, not donor dependency.

Weighing and measurement remain sites of systematic theft from the poor. In agricultural markets, mandi laborers apply 'practiced and timely nudges to the scale,' extracting 1-3 kg per quintal. In retail, informal sellers use rigged scales. The poor bear the cost because they cannot afford calibration equipment or have no recourse.

A smartphone-based scale verification system using computer vision and accelerometer data that allows any farmer or consumer to verify weight accuracy within seconds. Paired with a reputation system that publicly scores merchants on weighing accuracy, creating market incentives for honesty.

Community health workers in low-resource settings are overburdened with forms and protocols, spending up to 40% of their time on paperwork rather than care. They lack diagnostic support and operate in 'media dark' areas without connectivity. AI health tools exist but require reliable internet and assume trained medical professionals as users.

An offline-capable AI decision support tool for community health workers, designed for the specific constraints of rural practice: voice-based input in local languages, triage protocols that work without lab results, and sync-when-possible data collection. The system must assume intermittent power, limited literacy, and no specialist backup within 50 kilometers.

Passbook and bill design in financial services still ignores the material conditions of poverty. When Casas Bahia switched to a computer-generated passbook that didn't fit in a shirt pocket, default rates spiked—not because customers became dishonest but because they forgot. Physical reminders matter when you have no calendar app.

A 'BOP-native' design consultancy that audits financial products, housing programs, and service delivery systems for these last-meter failures. We would employ ethnographers and designers who have lived in poverty, not just studied it, to identify where the interface between institution and customer breaks down.

Self-Help Groups have proven transformative for women's economic empowerment, but their formation and training remain slow and dependent on NGO facilitators. Members adopted 'a certain color and style of sari to demonstrate their solidarity'—the social technology works, but it doesn't scale.

An AI-powered SHG formation and facilitation system that identifies potential groups from mobile money transaction patterns, provides audio-based training in local languages, and connects groups to formal financial institutions once they demonstrate savings discipline. The system would learn from successful group dynamics to predict which formations will sustain.

The 'last step' problem in water distribution—purified water parceled out in unhygienic containers and touched by unclean hands—negates the benefits of water purification systems. NGOs install purifiers; contamination happens between the tap and the mouth.

A complete system design that includes the dispensing and immediate-storage mechanism, not just the purification technology. Specifically: household containers designed to prevent hand contact with water, with embedded indicators for contamination levels, sold through the same distribution networks as other household goods.

Informal waste pickers provide 50-80% of recycling services in developing-world cities but receive no recognition, no safety equipment, no price transparency, and no collective bargaining power. They are the largest invisible workforce in the circular economy.

A platform that aggregates informal recyclers into verified networks, provides real-time pricing for different materials from multiple buyers, offers group purchasing for safety equipment, and creates verifiable credentials that formal waste management companies can recognize. The system would be owned by recycler cooperatives, not by a platform company extracting rent.

Housing construction for the poor proceeds without technical expertise, resulting in material waste, structural defects, and rooms built without planning. Most families employed local semiskilled or unskilled masons who built rooms without any planning.

An AI construction advisor delivered through WhatsApp or voice calls that provides step-by-step guidance for room additions: material calculations to minimize waste, structural advice for local conditions, and quality checkpoints with photo verification. Integrated with materials financing so that technical advice and credit arrive together.

Government officials in developing countries enter 'fictitious' data into monitoring systems to meet reporting deadlines, then scramble to present actual numbers when scrutinized. Real-time dashboards create accountability theater rather than actual performance improvement.

A verification layer for government monitoring systems that cross-references official data with independent sources—satellite imagery for infrastructure claims, mobile phone activity patterns for population movement, randomized citizen feedback via SMS. The system would flag discrepancies for automatic escalation, making data fabrication riskier than honest reporting of poor performance.

Credit analysts serving BOP customers use informal signals—'if a customer comes in and says he is a construction worker, the analyst will notice if the customer has calluses on his hand'—that AI credit scoring systems cannot replicate. As financial institutions scale, they replace this human judgment with formal data requirements that exclude the honest poor.

An AI credit training system that captures and teaches the tacit knowledge of experienced BOP credit analysts. Using video recordings of actual customer interactions (with consent), the system would identify the subtle signals that predict repayment—physical indicators of claimed occupations, conversational patterns that indicate honesty, community reputation signals—and train new analysts to recognize them.