The Infrastructure We Keep Promising Ourselves

Why the humanitarian reset needs shared digital public infrastructure, and an invitation to help build it

Every crisis leader remembers a moment when a decision had to be made before the picture was complete. In my own career, those moments rarely came because information did not exist. The municipality knew which water systems had failed. A local organization had already walked the affected villages. A responder had a spreadsheet of what had been delivered and where. People in the community were describing, in their own words, exactly what they needed. The information was there. It simply never came together in the room where the decision was being made, and it almost never found its way back to the people who had provided it.

I have come to believe that this is not a story about individual crises. It is a story about the architecture of the humanitarian system itself, and it has become impossible to ignore.

A system at an inflection point

The numbers are sobering. International humanitarian assistance fell to US$33.3 billion in 2025, the lowest level in a decade, and UN-coordinated appeals were only 35 percent funded. Funding per person targeted has roughly halved since 2021. The Humanitarian Reset, launched in March 2025, has asked the sector to become lighter, faster, more local and more accountable. ALNAP’s new State of the Humanitarian System report captures the mood well: practitioners see the diagnosis of the Reset as broadly correct, but the response as insufficient.

What strikes me most is that we may be living through two resets at once. The first is an efficiency reset, and it is well underway: fewer clusters, sharper prioritization, consolidated services. The second is the redistribution of power that the sector has promised for a decade, and it is moving far more slowly. Local and national organizations make up the large majority of humanitarian organizations, yet they received only about 4 percent of humanitarian funding directly in 2025, against a Grand Bargain target of 25 percent agreed in 2016.

When intentions are this widely shared and progress is this slow, the problem is rarely a lack of goodwill. It is usually a matter of incentives, and of the infrastructure that shapes them.

The layer nobody owns

Consider what we ask of a twenty-five person local organization in a disaster-prone country. We ask it to demonstrate governance, financial controls, safeguarding and data protection to the standard of a large international agency. We then ask it to do so separately for every partner, often through assessments that overlap almost entirely. One study of due diligence in Ukraine estimated about 80 percent overlap in the content international NGOs asked local partners to provide. Meanwhile, the program grants that fund its work rarely pay for the accounting system, the cybersecurity or the data tools that such accountability requires. We ask for transparency and then decline to fund the infrastructure of transparency.

Consider, too, what happens to shared goods in a funding shock. In 2025 several of the services that the whole sector relied upon for data and analysis were shut down or scaled back when single donors withdrew. Open crisis data that was available and up to date fell from 74 to 68 percent across the operations tracked by OCHA’s Centre for Humanitarian Data. We had been financing public goods as if they were private projects, and we discovered how fragile that made them.

And then there is artificial intelligence. Surveys suggest that more than nine in ten humanitarian workers have used AI tools, while fewer than one in ten organizations report that AI is integrated in a governed way. AI does not replace leaders. It exposes them. It is now exposing a sector that never built the shared foundations, the common standards, the trusted data and the governance, that would allow AI to strengthen local capacity rather than deepen old dependencies.

An idea worth testing

Over the past months I have been working on an idea I call HumanitarianOS. It is a proposal for independent, locally governed digital public infrastructure for humanitarian action. The name may suggest software, but the idea is closer to a road network than an application. It asks a simple question: what if the digital capabilities that reform depends on were treated as shared public infrastructure, instead of something every organization must build and finance alone?

In practice, that means three things working together. The first is a thin, open layer of standards, trust and exchange that connects the systems we already have, from HDX and KoboToolbox to the IFRC GO platform, the UN Partner Portal and national disaster information systems, rather than replacing them.

The second is a Shared Operational Picture, which brings together what governments, local organizations, international responders and communities know about a crisis, shows clearly what is verified, contested or unknown, and returns verified information and actions to those who contributed it.

The third is an Organizational Layer that gives local organizations a portable way to prove their trustworthiness once and share that evidence under their own control, alongside modern tools for finance, grants and reporting that they would never otherwise be able to afford.

AI sits across all of it, governed carefully, assembling evidence and reducing paperwork while people keep authority over the decisions that matter.

What excites me most is what happens when these pieces meet. If an organization records its activities once, and that information can flow with permission into a shared picture of who is doing what and where, coordination stops being a separate reporting burden and becomes a by-product of ordinary work. Duplication can be spotted before a program is designed rather than after it is delivered.

None of this works without independence. Infrastructure that one donor, one agency or one technology company controls will never earn the trust of everyone else. The governance has to be designed in from the first day: local and national actors holding real voting power, no controlling vote for any single institution, money that buys no votes on standards, data that stays where it is generated, and independent audits that are published. Trust is not a feature of shared infrastructure. It is the foundation on which everything else stands.

Where I am starting, and why I need you

I have learned to distrust grand designs that arrive fully formed, so I am proposing to begin modestly. The first step is a three-month period of visioning and partnership building, focused on listening before building. The aim is to test whether local and national organizations, governments and the wider humanitarian community want this infrastructure, and whether they are willing to help govern it. If the answer is yes, an eighteen-month phase would build the core in the open, using open-source tools, and pilot it in three countries alongside the organizations who will use it, with independent evaluation from the start. If the answer is no, that is valuable knowledge too, and the work should stop.

This is not something one person can do, and it is not something that should be designed far from the places where crises happen. So this is an invitation.

If you lead or work in a local or national humanitarian organization, I would like to learn where the burdens of compliance, reporting and coordination weigh most heavily on you, and what would make shared infrastructure worth your trust.

If you work in a government disaster management agency, a UN agency, the Red Cross and Red Crescent Movement or an international NGO, I would like to explore how this could connect to and strengthen the systems you already run.

If you are part of the crisis informatics, digital public infrastructure or responsible AI communities, I would value your help in designing the evaluation, the standards and the safeguards, and in challenging the assumptions behind this idea.

If you are a funder or work in technology, I would welcome a conversation about how to invest in a neutral public good that no single organization owns, including your own.

The humanitarian system is being asked to do more with less, at a time when the needs are growing and the old certainties are gone. Leadership in moments like this is less about authority and more about responsibility: the willingness to build the shared foundations that others can stand on, even when the credit cannot be claimed by any single institution. Resilience is a strategic advantage, and shared infrastructure may be one of the most practical ways for a contracting system to become a more resilient one.