.png)
Humanitarian analysis is a critical process in understanding and addressing the needs and challenges faced by communities in crisis. With the goal to ensure that aid is delivered in a way that is timely, relevant, and responsive to the needs of affected populations, humanitarian analysis seeks to inform decision-making in complex, high-pressure environments. By using diverse tools and methodologies, it helps practitioners anticipate future needs, reduce risks, and deliver aid that makes a meaningful difference in the lives of those who need it most.
The use of Artificial Intelligence (AI) in humanitarian analysis has revolutionized the way data is collected, processed, and utilized to respond to crises. AI technologies enable the rapid analysis of vast amounts of data from diverse sources, allowing humanitarian organizations to gain real-time insights into emerging crises, identify patterns, and predict future trends with greater accuracy.
Today we kick off the Human Algorithm series to introduce you to the people behind our data systems and showcase their knowledge. We are starting with José C. Cobos Romero, Head of Partnerships and Innovation at Data Friendly Space (DFS). José is a data and humanitarian expert with extensive experience in analytics, joint analysis, and information management for disaster response. As the Head of Analysis, Partnerships, and Innovation at DFS, José leads initiatives aimed at improving data accessibility and crisis response through AI. He focuses on developing innovative tools, workflows, and partnerships to support evidence-based decision-making during emergencies.
With him, we discussed what are the challenges an analyst faces, and the opportunities for further AI development in the humanitarian field.
Can you describe a typical day in your role as a humanitarian analyst?
Our typical day has transformed dramatically since we developed the GANNET tools. Before, we'd spend most of our time on mechanical tasks – searching for relevant documents, downloading PDFs, reading through reports to find key information, manually tagging content, cleaning datasets, and trying to organize information in a usable way. The actual analysis work was squeezed into whatever time remained.
Now, GANNET's knowledge database has completely changed this balance. When we start our day, we log into the Virtual Assistant for sudden onset emergencies or SituationHub for established crises. The system has already ingested sources overnight – reports from agencies on the ground, media coverage in multiple languages, social media trends, and official bulletins. Rather than hunting for information, we can immediately query the system about specific situations or areas of concern. The Virtual Assistant retrieves relevant information with sources clearly indicated, saving hours of search time. This shift means we now spend much more of our time on what our brains do best – contextual analysis, pattern recognition across seemingly unrelated data points, and bringing human judgment to complex situations.
With SituationHub specifically, we're doing this curation and verification work so that others don't have to. Once we validate the AI-generated analysis, external analysts in-country, decision-makers, civil society workers, and advocacy professionals can directly access reliable, up-to-date information without duplicating the same labor-intensive process. They can immediately use these verified insights for operational planning, funding decisions, or advocacy campaigns without spending days gathering and processing the same information themselves. For example, when working on the Sudan SituationHub, our analysis and verification work enables humanitarian coordinators to quickly understand cross-sectoral needs, local NGOs to identify priority areas for intervention, and donors to make evidence-based funding decisions – all from the same verified knowledge base that we've curated.
Working with AI hasn't made our roles less important – it's actually made our human skills more valuable. The machine handles the volume, but we provide the meaning and context that transforms information into understanding. And through SituationHub, this understanding becomes immediately accessible to everyone involved in the response, dramatically improving coordination and efficiency across the entire humanitarian ecosystem.
How do political instability or conflict in regions you analyze impact your ability to gather accurate data?
Political instability and conflict create significant challenges for accurate data collection. When violence restricts physical access to affected areas, the entire information landscape becomes compromised. In active conflict zones like Sudan, we face numerous data challenges - from limited field assessments to deliberate information manipulation by conflict parties. When primary data collection becomes impossible due to security concerns, secondary data analysis becomes absolutely critical. It's our duty to provide the most accurate, verified, and triangulated information possible in these challenging environments. This is precisely why we've designed our GANNET platforms with robust verification mechanisms.
For SituationHub, we tackle this challenge by integrating multiple information sources and making the process transparent. Users can see exactly which sources informed specific analysis points, allowing them to evaluate the reliability themselves. We deliberately include diverse source types - from international organizations' reports to local media and academic research - to provide a more complete picture when direct access isn't possible.
The Virtual Assistant takes this further by enabling users to check citations specific to each paragraph and trace information back to original sources. This source transparency is crucial when working with secondary data in contested environments, where information may be politicized or manipulated.
We've also found that temporal analysis helps overcome data limitations - tracking how reporting changes over time can reveal patterns and inconsistencies that point to information gaps or potential bias. By maintaining comprehensive archives of reporting on a crisis, we can identify when new information contradicts previous accounts and flag these discrepancies for further verification.
Another approach we use is clearly communicating confidence levels in our analysis. We're transparent about information gaps and the limitations of available data, which is ethically essential when informing humanitarian decisions in politically complex environments.
Ultimately, while conflict severely constrains data collection, our combination of comprehensive source integration, transparent verification processes, and human analytical judgment allows us to provide the most reliable picture possible under difficult circumstances. The feedback mechanisms built into both SituationHub and the Virtual Assistant are constantly improving this process, as users help identify information gaps and provide additional context that enhances the overall analysis.
How do you manage the challenge of balancing immediate humanitarian needs with long-term development goals in your analyses?
We constantly navigate the tension between addressing immediate humanitarian needs and supporting long-term development goals in our analytical work. This is one of the most persistent challenges in the sector, often referred to as the humanitarian-development nexus.
With GANNET SituationHub, we've designed our analytical frameworks to capture both dimensions simultaneously. What might appear as a straightforward situation report is actually built on a robust analytical structure that examines multiple layers of reality - from underlying drivers and root causes to immediate physical harm and humanitarian conditions. For example, in the Sudan SituationHub, our framework deliberately examines causal relationships between political instability, resource competition, and infrastructure breakdown, while also documenting their immediate humanitarian consequences. This structured approach helps us understand not just what's happening, but why it's happening and how different aspects of the crisis interconnect.
Critically, we ensure our analysis directly informs decision-making processes. We structure our outputs specifically to support response planning and programming decisions. This means organizing information in ways that help humanitarian actors identify priority interventions, allocate resources effectively, and design programs that address both immediate needs and underlying vulnerabilities. Our analysis isn't just about understanding the situation – it's about providing actionable insights that enable better humanitarian outcomes.
AI helps us process vast amounts of information across these different pillars of analysis, but our analysts make the critical connections between short-term interventions and their potential long-term implications. By examining everything from conflict drivers to humanitarian access constraints within a consistent framework, we can show how immediate response efforts might either reinforce or undermine longer-term resilience.
We've found that temporal analysis is particularly important - tracking how situations evolve helps identify patterns that might indicate where immediate response could either support or hinder development trajectories. SituationHub's regular reporting cycle (weekly and monthly analyses) creates a valuable historical record that reveals these longer-term trends.
Another practical approach we've adopted is structuring our reports to explicitly link immediate needs to their root causes and potential sustainable solutions. Rather than simply describing current conditions, we contextualize them within broader social, economic, and environmental systems. This helps response actors design interventions that address urgent priorities while considering how they might transition toward sustainable development approaches.
The reality is that there is rarely a perfect balance - each context requires unique considerations. But by ensuring our analysis captures multiple timeframes and system dynamics, we give humanitarian decision-makers the information they need to make more nuanced choices about how to meet immediate needs while supporting longer-term resilience and development.
What are some key opportunities for further AI development in the humanitarian field that you're excited about?
We're particularly excited about the shift from reactive to anticipatory approaches in humanitarian work. One of the most promising opportunities we see is leveraging AI to anticipate crises before they fully materialize, rather than just responding to them after they've occurred.
With GANNET's ability to process and compare millions of pieces of information across time and contexts, we're beginning to identify early warning indicators that traditional analysis might miss. By analyzing patterns from historical crises and comparing them with real-time data, we can spot concerning trends much earlier in their development. For example, we're exploring how to train models that can detect subtle signals across multiple dimensions - environmental changes, conflict dynamics, population movements, market fluctuations, and media sentiment - to forecast potential humanitarian needs weeks or even months before they reach critical levels. This moves us firmly into the realm of Anticipatory Action (AA), where we can help organizations preposition resources, release funding, and initiate preparedness measures before a crisis fully manifests.
The Anticipatory Action Center we've included in our SituationHub roadmap reflects this vision. We're working to develop predictive analytics capabilities that can forecast risks with increasing accuracy and trigger preemptive measures. This approach has the potential to fundamentally transform humanitarian response - making it more proactive, cost-effective, and ultimately reducing human suffering by addressing needs before they become acute.
What makes this particularly exciting is that we're not just theorizing, we already have the data infrastructure and AI capabilities in place with our existing platforms. The next step is refining these forecasting models, which requires close collaboration between our technical teams, analysts, and humanitarian partners to ensure the predictions translate into meaningful early action.
Moving toward anticipatory approaches means we can potentially shift resources from emergency response to prevention and mitigation, which is not only more efficient financially but also reduces the overall impact of crises on affected populations. We see this as the future of humanitarian work, and we're enthusiastic about the role AI can play in making this transition possible.
Continue to follow along as we dive deeper into the Human Algorithm.
