Louis Potter
Researcher
EPFL EssentialTech

With:
Anna Leander, Louis Potter, Rachel Howell, Silke Oldenburg, Nora Doukkali and Edgard David Rincón Quijano

Doing more with less: our starting point

• 305 million people projected to need humanitarian assistance in 2025.¹OCHA, Global Humanitarian Overview 2025 (United Nations Office for the Coordination of Humanitarian Affairs, 2025).
• A projected shortfall of 11 million health workers in low and lower-middle income countries by 2030.²Global Strategy on Human Resources for Health: Workforce 2030 (World Health Organization)
• Donated medical equipment routinely idle within months when delivered without co-planning, training, and spare parts.³Perry, L., and Malkin, R., ‘Effectiveness of Medical Equipment Donations to Improve Health Systems’, Medical & Biological Engineering & Computing 49(7), 2011; Marks, I. H. et al., ‘Medical Equipment Donation in Low-Resource Settings’, BMJ Global Health 4(5), 2019.

1.
The missing connection in complex knowledge ecosystems

International humanitarian organisations (IHOs) are responding to growing complexity with tighter resources, fewer specialist staff, and increasing pressure to maintain quality across dispersed operations. Field teams need quick, practical support. Technical referents at headquarters are stretched across multiple countries and requests simultaneously.

A technical issue in a health setting rarely sits in isolation. Equipment problems involve maintenance, power, consumables, staffing, training, and procurement. A report of rising malnutrition may point to drought, water access, or food insecurity rather than a clinical gap. The visible question sits in one sector; the root cause may sit across several. No one is a specialist in everything, and the cross-domain translation work—connecting water to clinical care, nutrition to surveillance, infrastructure to training—currently lives in high-level management documentation or the minds of managers.

HUD’s scoping research, drawing on desk research, key informant interviews, and a decade of field experience, has traced these patterns across contexts. The same challenges recur: rushed device selections, knowledge loss through staff turnover, weak planning for project closure, and data handover gaps. “Graveyards of medical equipment… Very high-end equipment… the maintenance cannot be maintained,” one IHO specialist reported. These failures have a common root: a lack of connection between specialist knowledge at the moment decisions are made in these interacting systems.

2.
From building knowledge to connecting knowledge

Specialist knowledge platforms already exist for most humanitarian fields—water and sanitation, clinical medicine, food security, global health guidelines, and more. Dashboards of the different clusters also contain publicly available information that is useful but siloed in a way that makes it unusual for specialists to see the whole picture. These platforms are well-curated and actively maintained. Building another hub would mean duplicating their work and competing with them for the same contributors.

What is missing is a way to navigate between them and delivering solutions that are specific to local contexts, requirements and personas.  A connector that takes a user’s question, understands their role and context, routes the query to the most appropriate specialist platform—or several, when the problem crosses specialisms—retrieves relevant guidance, and adapts the answer to the person asking. The specialist platforms keep their content. The connector does the transversal-work.

On the HQ side, the same interactions produce a live picture. Instead of each field request remaining an isolated exchange, recurring themes become visible across sites and contexts. If several country offices independently report similar failures, the pattern surfaces—pointing to a shared supplier issue, an infrastructure gap, or a training need that no single request would reveal on its own.

3.
How it works

The connector is designed around the person asking. When a user poses a question, the tool considers who they are and where they are working: their role, their location, the language they are most comfortable in, and the specific guidelines and protocols that apply in their context. A clinical officer in a national health facility and a community health volunteer in a refugee camp may face the same underlying issue, but the guidance they need—and the way it should be framed—can be very different. The connector tailors its response accordingly, drawing on the right specialist platforms and adapting what it finds to the user’s operational reality.

This also begins to address a longstanding information imbalance between the field and headquarters. Local healthcare workers and frontline staff are often the first to encounter emerging problems, but their questions and observations rarely feed back into organisational decision-making in a structured way. The connector creates a channel in both directions: field teams get timely, contextualised support, while HQ gains an ongoing stream of data showing what people are asking, where, and how often. Patterns that would otherwise surface only through formal reporting—or not at all—become visible earlier, allowing problems to be addressed before they escalate.

Both scenarios below cross several specialisms. Both work the same way.

CHOLERA EXAMPLE
A WASH practitioner in a refugee camp asks: “Are we seeing cholera here? What should I do?” The question crosses water, clinical care, public health, and food handling. The connector identifies the user’s role and location, routes across the relevant specialist platforms, and returns what to do in the next two hours, what to escalate, and what to record—all framed for a WASH practitioner working in that specific context. HQ sees the cluster pattern across other camps—supplies pre-mobilise without an explicit alert.

ACUTE MALNUTRITION
A community health volunteer reports: “MUAC red-bands rising for the third week—what now?” The question crosses clinical assessment, food security, surveillance, and water access. The tool returns referral thresholds, what to log, who to alert and when—adapted to the volunteer’s language and the local referral protocols in place. HQ sees coordinated regional rises across many sites—well before any single explicit escalation.

The Architecture

The connector sits as a middle layer between the people asking questions and the specialist platforms that hold the answers. At the top: anyone with a question—a field worker, a local clinician, an HQ planner—posing it in their own words. In the middle: the connector, which interprets the question, identifies the user’s role and operating context (location, language, applicable guidelines), selects the most relevant specialist platforms, retrieves guidance from them, and synthesises an answer adapted to the person asking. At the base: the specialist platforms themselves—a WASH knowledge base, a clinical and medical resource, a food security monitoring system, a guidelines repository, and others as the network grows. Each platform keeps its own content and curation. The connector does not duplicate or replace it; it navigates between them, and bridges across them when a question demands more than one specialism.

As a side-effect of every interaction, the connector quietly feeds patterns to HQ—what is being asked, from where, and how often—turning isolated field questions into organisational intelligence.

4.
Where we are going

The connector model means building partnerships one at a time. The first step is a light proof-of-concept with a single specialist partner, focused on two or three real cross-domain journeys—testing whether the tool can interpret a question, route it to the right platform, and adapt the answer for the person asking. A working pilot becomes the credibility for the next conversation, extending the network hub by hub.

In parallel, co-design workshops—building on work already underway with partners in Barranquilla, Colombia—helped ground the tool in real operational workflows (link through to Colombia reports when online). The humanitarian sector has seen a proliferation of digital tools that promise transformation but often replicate existing power asymmetries or further distance humanitarians from the people they serve. This project is grounded in the recognition that a connector can only be useful if shaped by the constraints and realities of the people who will use it.

The concept is already concrete enough to discuss seriously—wireframes and a development pathway exist—but still early enough for partners to influence what a first prototype should do, how it should be bounded, and how it should be tested. For organisations under pressure to do more with fewer resources, this is an opportunity to help shape a practical tool without asking anyone to rebuild what already works.