For economies across the Global South, innovation has become decisive. Facing overlapping shocks, from economic volatility and climate disruption to rapid demographic and technological change, their capacity to innovate increasingly shapes whether they can generate jobs, diversify production, strengthen the state and withstand crises. Yet time and again, promising innovations succeed as pilots and then stall, rarely reaching the scale or permanence they need.
That problem brought 194 researchers, policymakers, practitioners and partners from 33 countries to Nairobi, Kenya, on 12 June 2026 for PEP's annual Policy Conference, Rethinking the Future of Innovation: What It Takes for Systems to Deliver in the Global South.
As Prof. Jane Mariara, PEP's Executive Director, set out in her opening remarks, PEP supports Southern researchers working on the questions their own countries are asking, trains researchers and policymakers, and creates the spaces where they engage with one another. The conference followed this ethos, with Southern experts leading the discussion as it moved from why systems fail to what would make them work.
The system decides whether an innovation scales
An innovation succeeds or fails on the strength of the system around it. This was the argument of the keynote address, delivered by Wonhyuk Lim. Prosperity, he told the conference, comes from continuous waves of innovation, each one resting on capabilities, incentives and politics working together.
"STEM education is necessary but not sufficient. Without firms ready to hire graduates, talent leaves the country."
— Wonhyuk Lim, Associate Dean, KDI School of Public Policy and Management
Indeed, as Mamadou Biteye (African Capacity Building Foundation) said in his opening remarks, Africa has watched solutions piloted, praised and then abandoned once the money runs out, leaving little behind.
"The missing ingredient is the system around the idea, the institutions that govern it and the incentives that make that system work. Brilliant innovations without capable systems are just ideas."
— Mamadou Biteye, Executive Secretary, African Capacity Building Foundation
During the panel discussion on why innovation systems struggle to scale, panellist Muthoni Masinde explained how most innovations die in the gap between a successful pilot and routine institutional practice, a gap she called the “Valley of Death”. Fellow panellist Mónica Salazar described how these failures compound across Latin America, where weak coordination, scarce intermediaries and little evaluation reinforce one another until policymakers can no longer tell what works.
Systems overlook the innovation that already exists
Too often, systems fail to see the innovation in front of them. As Muthoni Masinde put it in the panel, the knowledge often already exists, held by the people closest to the problem, yet invisible to formal systems not built to hear it.
"Pilots often succeed because they are relational, community-driven and highly supported. Scaling fails when systems attempt to replace trust, local participation and contextual adaptation with bureaucratic standardisation."
— Muthoni Masinde, Dean, Faculty of Engineering, Built Environment, and Information Technology (FEBIT), Central University of Technology, Free State, and Founder of the ITIKI (Information Technology and Indigenous Knowledge with Intelligence) platform
Muthoni’s own work shows the cost. Across Kenya, South Africa and Mozambique, smallholder farmers detect environmental stress months before formal early-warning systems react. Her ITIKI platform, which integrates information, technology and indigenous knowledge so that community observation feeds into formal climate and planning systems, captures that intelligence. This kind of knowledge is innovation infrastructure, she argued, and treating it only as heritage to preserve wastes it.
The question of who counts as an innovator had been raised from the outset of the day. During the opening remarks, Eunice Muthengi (Mastercard Foundation) said that young women should help shape the research questions themselves, rather than appear only as the people to whom research is done. Kathryn Toure (International Development Research Centre) extended the point to how research is produced, arguing it holds up best when those it affects help build it, with capacity strengthening and inclusion designed in from the start. Similarly, Margaret Mliwa (Ford Foundation) made the point that communities hold the solutions to their own problems and as such funders do best when they walk alongside community partners rather than direct from above.
On the panel, Chinasa Okolo described this as a matter of epistemic justice. African knowledge systems, she argued, belong in the design and governance of technology, not only in its rollout. Tonny Omwansa, a fellow panellist, found the same exclusion in his own recruitment at the Kenya National Innovation Agency. The candidates who reach innovation roles skew male, urban and older, which he put down to a narrow sense of who belongs in innovation leadership rather than any shortage of capable people.
Audience members during the discussion gave concrete examples, from women talked over in technical spaces to young people, including those with disabilities, pushed to the margins of innovation rather than treated as a source of knowledge.
A shared conclusion emerged across the sessions. Systems that exclude these groups lose knowledge they cannot reproduce elsewhere, and they perform worse as a result.
Imported solutions rarely fit local realities
As well as overlooking the knowledge in front of them, systems also lose touch with local realities when they adopt solutions designed somewhere else. Further, innovations can create new development challenges if poorly governed, as Ingrid Woolard, member of the PEP Board of Directors, cautioned while opening the conference.
Artificial intelligence provides a clear example as Southern governments face pressure to adopt AI systems they did not design, cannot inspect and cannot explain to the communities affected. Chinasa warned that frameworks built elsewhere, such as the European Union's AI Act, do not transfer to Nigeria, Chile, Indonesia or Papua New Guinea, and that the remedy begins with the ability to see inside the systems being bought.
"You cannot govern what you cannot audit. Governments and funders must invest in accountability infrastructure, civil society AI auditors, parliamentary technology committees and community feedback mechanisms."
— Chinasa Okolo, Founder & Scientific Director, Technecultura
Rohinton Medhora, Chair of the PEP Board of Directors, shared that together, China and the United States hold as much as to 95% of the world's AI patents, so even a modest local procurement choice is bound up in global rivalries.
However, this does not argue for closing the door on outside knowledge. In the keynote discussion, Wonhyuk pointed to Japan, which built much of its modern economy by absorbing foreign knowledge and adapting it to its own conditions. He contended that knowledge travels, and it has to be translated for where it lands.
Institutions decide whether innovation lasts
Overlooking local knowledge, importing what does not fit, and failing to carry pilots into lasting practice are all failures of institutions. This raised the question: what does an institution that can absorb, adapt and sustain innovation actually need?
Tonny argued that leadership is the first requirement. Running processes and meeting compliance targets is a different skill from spotting an opportunity and mobilising people and resources behind it, and public institutions treat them as one and the same. He described a system that works against innovation, where performance contracts reward compliance, procurement is not built for new ideas, and officials who go beyond their brief earn little credit for it.
"The interest is there; what is missing is institutional permission and momentum."
— Tonny Omwansa, Chief Executive Officer, Kenya National Innovation Agency
Mónica pointed to the connective tissue: the intermediary bodies that link universities, business, government and communities are usually too few and too weak, and their absence is the tightest constraint on the whole system. Evaluation, she added, rarely closes the loop, since even assessed programmes seldom feed their findings back into decisions to adjust, drop or scale them.
"Policymakers frequently do not know what works, what does not, and why."
— Mónica Salazar, Managing Partner, Technopolis Group Colombia
What neither leadership nor coordination resolves is the pressure a government is under while it builds them, and that was the tension the afternoon simulation, facilitated by Rohinton, exposed. In mixed teams, participants advised on two cases, scaling a programme in a cash-strapped state and designing a youth-focused green innovation ecosystem, and were asked which trade-offs they would be willing to accept. Among the teams that reported back, the answer was consistent: they would protect quality and lasting impact over speed and coverage, accepting slower progress rather than rushing to scale. Unprompted, they had chosen to protect the participation and trust an innovation depends on: the very thing Muthoni had warned bureaucratic systems strip out.
Closing the conference, Alemayehu Seyoum Taffesse, member of the PEP Board of Directors, described an innovation system as a web of actors, rules and relationships that works only when those parts evolve together. For developing economies, this means treating innovation policy as adaptive and experimental, and repurposing the institutions already in place.
"What we have is not destiny; it is where we start."
— Alemayehu Seyoum Taffesse, Member of the PEP Board of Directors
Six priorities for action
The diagnosis pointed to a clear set of priorities, calling for policymakers, partners and researchers to:
- Finance the whole innovation journey, not only the pilot. Fund on a long enough horizon to carry promising work from pilot into routine institutional practice.
- Grow entrepreneurial leadership inside the public sector. Redesign performance frameworks and incentives to reward the people who turn ideas into delivery.
- Treat community and indigenous knowledge as innovation infrastructure. Build standing mechanisms to bring it into planning and service delivery.
- Develop AI governance with local actors. Build the capacity to buy AI critically, audit it actively and refuse systems that carry the wrong assumptions.
- Design inclusion in from the start. Use territorial and local approaches that treat marginalised groups as innovators and co-creators.
- Build the coordination layer. Establish intermediary bodies that hold policy steady across political cycles and connect academia, government, communities and markets.
About the event
The 2026 PEP Policy Conference took place on 12 June 2026 at the Holiday Inn in Nairobi, Kenya, as one day of PEP's 2026 Annual Conference. Proceedings were livestreamed. The conference, and the convening that made it possible, were organised with the support of our programme partners: the International Development Research Centre, the Mastercard Foundation, the William and Flora Hewlett Foundation, Global Affairs Canada, the Fonds d'Innovation pour le Développement, and the Global Education Analytics Institute.
Related posts
