Peer-Led HIV Care and Adaptive Learning Systems

Project Overview

The Yim’lo Project was a lived experience-led HIV service delivery programme supporting men who have sex with men and transgender women across three geographies in South Africa. The programme was led by Population Services International (PSI) South Africa, in partnership with The Aurum Institute, Matchboxology, and Pivot Collective.

At its core, Yim’lo centred a peer-led model, in which trained peer coaches provided tailored support to individuals navigating HIV prevention, testing, and treatment services. Grounded in shared lived experience, the model aimed to build trust, reduce stigma, and strengthen engagement with health services among communities often underserved by traditional systems. Coaches played a critical role as bridges between communities and the health system, supporting individuals to access services, navigate disclosure, and advocate for their health needs. The programme emphasised dignity, empathy, and non-judgemental engagement as foundational to effective service delivery.

Our Role

Pivot Collective partnered on the Yim’lo Project as the learning and evaluation partner, designing and stewarding an adaptive learning approach that enabled the programme to evolve in response to real-time insights from implementation.

Our work was grounded in the principle that effective service delivery models, particularly those working within complex and shifting social contexts, require ongoing cycles of reflection, learning, and adaptation. Rather than positioning evaluation as a retrospective or external process, we embedded learning within implementation, creating structures for continuous feedback and programme refinement.

This adaptive approach was operationalised through a co-design and co-learning process, recognising coaches and implementers as central contributors to knowledge generation. We invested in building the capacity of coaches as peer researchers, developing tailored training curricula and facilitating participatory workshops focused on qualitative and participatory research methods. Through this, coaches were able to engage directly in data collection, reflect on their own practice, and contribute to interpreting emerging insights.

Working alongside PSI, Aurum, and other partners, we facilitated iterative learning cycles across the pilot, including mixed-methods data collection, rapid analysis, and participatory sense-making processes. These cycles created structured opportunities for coaches, service providers, and programme teams to collectively interpret findings and translate them into concrete adaptations.

This approach ensured that learning was not extracted from the programme, but co-produced within it, enabling timely adjustments to key components of the model, including coach support systems, digital tools, and participant pathways.

The process culminated in the co-development and authorship of an adaptive implementation guide, which captures the core components of the model, key learning from the pilot, and practical guidance to support future adaptation and scale across different contexts.

What We Learned

Learning from the Yim’lo pilot highlights that peer-led models of care operate as deeply relational systems shaped by trust, identity, and lived experience.

At the centre of the model is the role of the peer coach. Coaches do far more than facilitate access to services. They build relationships grounded in shared experience, empathy, and non-judgement. This relational foundation enables forms of engagement that are often difficult to achieve through formal health systems alone, particularly for men who have sex with men and transgender women navigating stigma and exclusion. The effectiveness of the model depends not only on what coaches do, but on who they are and how they are supported to show up in these roles.

This has important implications for how peer-led programmes are designed and resourced. The pilot highlighted the extent to which coaches carry significant emotional and relational labour, often navigating complex situations alongside the individuals they support. Sustaining this work requires structured investment in mentorship, psychosocial support, and collective spaces for reflection. Without this, the very qualities that make peer models effective can also place strain on those delivering them.

The project also surfaced the importance of identity and representation within peer models. Shared identity, whether in terms of gender, lived experience, or social context, plays a critical role in building trust and enabling meaningful engagement. At the same time, the diversity within communities means that no single model of “peer” is sufficient, reinforcing the need for flexibility and responsiveness in programme design.

From a systems perspective, the pilot demonstrated the value of embedding adaptive learning within implementation. Rather than treating evaluation as a retrospective exercise, iterative cycles of data collection, rapid analysis, and participatory sense-making enabled the programme to respond to emerging challenges and opportunities in real time. This shifted evaluation from a tool of accountability alone to a mechanism for ongoing programme improvement and shared ownership across partners.

Finally, the development of the implementation guide reinforced that scaling such models is not about replication, but about adaptation. What matters is not reproducing a fixed approach, but carrying forward its underlying principles (relational care, community leadership, and responsiveness to context) into new settings. The Yim’lo pilot highlights that effective scale depends on maintaining these principles while allowing the model to evolve in response to the realities of different communities and health systems.