1. The programme in brief
AI Diffusion Pathways for Agriculture (AIDPA) is a six-month programme by the AI Hub for Sustainable Development (co-led by the Ministry of Enterprises and Made in Italy and the United Nations Development Programme), with DeepLeaf as delivery and platform partner.
The programme addresses a persistent challenge: agricultural knowledge exists in abundance, yet often fails to reach farmers when and where they need it.
Advances in AI and digital infrastructure offer new ways to bridge this gap. However, local agritech ventures often lack the capital, technical capacity and trusted local data needed to build reliable services. Without field partners and routes to customers, promising pilots struggle to become widely used products and viable businesses. As a result, farmers continue to make planting, treatment and input decisions without the timely expert advice these ventures could provide.
AIDPA helps ventures overcome these barriers and develop services that address real farming needs, are accessible to farmers and have viable business models. It treats AI as knowledge infrastructure: a shared capability that turns trusted agricultural research, locally relevant data and agronomic expertise into practical guidance. This approach connects models, technology and compute from G7 countries with validated local knowledge, enabling ventures to build services suited to their markets.
The programme will support seven agritech ventures - one each in Ghana, Italy, Kenya, Morocco, Rwanda, Tanzania and Tunisia - with shared infrastructure, technical and business support, and local partnerships. Each venture will adapt the shared AI platform to its crops, languages and communities, then test its service with real users.
Together, their experiences will show which capabilities, methods and partnerships can be reused and which require local adaptation. These lessons will contribute to the AI Hub’s ambition of 100 AI Diffusion Pathways by 2030.
2. The opportunity for ventures
From zero to one in six months: from technical build to adoption
Selected ventures receive support to develop an early-stage idea or grow an existing business. Shared AI infrastructure, validated local knowledge and local partnerships help teams build services tailored to their crops, languages and farming conditions, without developing their own underlying AI model.
01
Cohort workshops
Practical sessions with the full cohort, covering technology, product development and reaching customers.
02
Venture workshops
Sessions tailored to each venture’s technical, business and local market needs.
03
Technical assessments
Support from DeepLeaf to build, integrate and adapt the venture’s product.
04
Progress check-ups
Regular reviews of business and pilot progress to identify challenges and guide next steps.
05
Platform access
Free access to DeepLeaf’s harness, API and SDK until 31 July 2027, with a dedicated workspace and the venture’s own branding.
06
Adoption through pilots
Access to pilot sites, real users and customer introductions, supported by local partners and validated agricultural knowledge.
Intended outcomes
By the end of the programme, each venture is expected to have:
A working product and a completed pilot with real users.
A signed purchase or service contract with a customer or adopting institution.
A pitch prepared for the proposed Demo Day on 3 August 2027.
Intended outcomes
Ventures retain ownership of their products, customer relationships, interfaces, local operational data and business models, subject to applicable contracts and law. Portability is a programme principle, enabling ventures to continue developing their services beyond the cohort without being locked into a single provider.
3. Who can apply
La Banque africaine de développement (BAD), membre fondateur du comité de pilotage exécutif du AI Hub, et le AI Hub du PNUD pour le développement durable s'associent à Google et Apolitical pour renforcer les compétences en intelligence artificielle dans le secteur public. Cette initiative propose une formation gratuite de classe mondiale destinée aux agents publics, visant à combler les lacunes en matière de compétences et à renforcer les partenariats public-privé dans les domaines de l'infrastructure et de l'innovation liées à l'IA.
Ce programme s'inscrit dans le cadre du Apolitical Government AI Campus, un pôle mondial de référence pour l'excellence en matière d'IA au sein des gouvernements, qui prépare un million d'agents publics à devenir des leaders à l'ère de l'IA, avec le soutien de Google.org et de la Fondation Rockefeller. Le Campus est neutre sur le plan technologique et accompagne les dirigeants, les praticiens et les apprenants en phase de sensibilisation afin de leur permettre de comprendre l'IA, de l'appliquer de manière responsable et de favoriser des améliorations concrètes dans la prestation des services publics.
