Case Study | Claude Enterprise
The International Rescue Committee turns fragmented health data into faster decisions with Claude
The International Rescue Committee (IRC) delivers health care and humanitarian aid in some of the hardest places to work, serving 16.8 million people annually through more than 2,700 health facilities in 40 countries. At that scale, timely, practical data is essential to helping teams adopt programs, target outreach and plan staffing and supplies. Much of this information originates in national health systems and is captured across different paper and digital formats, creating an opportunity to make it easier for teams to bring data together, identify trends, and act on insights more quickly.
With Claude, the IRC built:
- Offline HTML dashboards covering more than 15 core health and reproductive health indicators, with no changes to how Ministry of Health facilities collect data
- A drag-and-drop workflow where staff load facility Excel files locally, no connectivity or coding required
- Time savings for every surveyed pilot user, with half saving four or more hours per month
- Data analysis that surfaced a sharp rise in STI consultations among under-18s, validating the team’s youth outreach efforts.
- Visibility into a previously invisible gap between prenatal care and facility births, prompting a field investigation
- Multi-month trend analysis that caught an early respiratory infection spike, informing staffing and supply planning
The challenge
Health data trapped in paper notebooks
Data-driven decision-making is central to how the IRC operates, particularly in its health programs. The organization relies on programmatic data to know when to shift a program model, where to focus outreach, and how to plan staffing and supplies. But most of the health facilities IRC supports are run by national ministries of health, which means the IRC can’t control what data gets collected or how. Formats differ from country to country and even between sites in the same region.
In much of sub-Saharan Africa, that data starts life in a paper register at the facility. Sometimes the numbers land in DHIS-2, a digital health system. But more often they stay on paper until IRC staff transfer the aggregate numbers into Excel by hand, a process that is time consuming. “With health programs, there’s a lot of seasonality to it, or windows for intervention periods close,” said Kristy Crabtree, global lead of technology for programs at the IRC. “You might catch something, but it’s too late to do anything about it.”
Standardizing the data was never a realistic option. “The scale of this problem is significant,” Crabtree explained. “It’s close to 2000 healthcare facilities in sub-Saharan Africa that have the same problem. Claude created an opportunity for us to build flexible tools, helping teams make better use of the data already available to them.”
The solution
An offline dashboard built in the chat window
Rather than overhaul the data pipeline, the team changed a single variable: visualization. In 2025, the IRC piloted a new approach to health data visualization, deploying Claude Enterprise to frontline health staff and measurement specialists. Burkina Faso served as one of the pilots, supported by an in-kind license from Anthropic. Staff load aggregate facility data, not individual patient records, into the dashboard locally. No connectivity or coding is required, and the data is not sent externally or stored on Claude’s servers.
“We went into this truly as an experiment,” Crabtree said. “Can Claude efficiently generate high quality visualizations of this data that’s in these inconsistent formats that we’re taking from tally sheets? Will people trust these visualizations? And are they going to bring these into their decision-making around programs?”
Working with the Burkina Faso team’s list of 15 core indicators, IRC used Claude to create a customized visualization template rendered as a single HTML file. That format choice solved two problems at once. Building and customizing an HTML dashboard the traditional way “would have required engineers involved in the process,” Crabtree said. “It would have been costly, and given the number of facilities that would need unique visualizations, we just wouldn’t have gone down that road.” Claude also let the team prototype and adapt the tool directly, at a scale that would otherwise have been out of reach.
Most importantly, it works where the internet doesn’t. “It’s a super light file that I can send to a facility manager and say, here’s the file,” Crabtree said. “Then they can take their Excel file from that local facility and just drag and drop it in there. No connectivity needed.”
IRC Measurement Advisor Banny Attiey serves as the local measurement focal point in Burkina Faso, updating the dashboard in natural language as health facility teams request changes. “The team was able to make the dashboard “speak” and steer the data presentation in the desired direction without needing any special expertise; it’s a step forward in real-time support for decision making.”
“The team was able to make the dashboard “speak” and steer the data presentation in the desired direction without needing any special expertise; it’s a step forward in real-time support for decision making.”
The outcome
Data reviews that turn into decisions
The shift showed up immediately in how teams worked. “We saw really quickly, even in their first data review meeting, how health care workers went pretty immediately into, ‘how can we think about this strategically’,” Crabtree said. Strategic planning for the health team in the pilot is now informed by all core indicators rather than the two or three indicators the teams previously worked from.
In pilot feedback surveys, every health and measurement respondent reported working more efficiently: half are saving four or more hours per month, and another 30% are saving two to three hours. One program user now prepares analysis and slides for senior management in under 30 minutes, work that previously took a full day. A senior program coordinator said portfolio-level synthesis that once required more than a month of manual Excel work is now immediately available.
The improved visibility is already influencing how teams investigate trends and plan services. Across three health facilities, the dashboard flagged STI consultations among people under 18 jumping from an average of 21 per month to 88, more than four times higher, since April 2025. The health team investigated the trend and concluded that awareness of the team’s youth-friendly services—and access to them—had increased. As a result, young people who previously would not have sought care were coming forward. “Rather than this being a cause for alarm, they were able to confirm from that data that their modified approach on outreach to youth was having a positive impact,” Crabtree said.
The dashboard also surfaced a gap no one had seen across facilities before: from January through May 2026, more than 6,000 pregnant women attended prenatal checkups at IRC-supported facilities, while only 1,762 births in the same areas were attended by medical personnel. A field investigation revealed how women in conflict-affected areas navigate pregnancy and childbirth, from traveling to relatives in safer locations to relying on trained community midwives when health centers close, findings now informing service access, planning, and outreach.
Another outcome of the dashboard was multi-month trend analysis. This used to be impossible when data was reviewed one month at a time. As a result of the analysis, the team caught a February 2026 spike in respiratory infections, more than 25,000 pneumonia cases in a single month, and traced it to unusually warm, dusty harmattan winds arriving ahead of the usual seasonal peaks. The team can now anticipate surges and plan staffing, supplies, and outreach accordingly.
“One of the most exciting parts of this is that it’s going to let us turn data we already had but couldn’t visualize into something we can actually use to make the decisions that we want to make,” Crabtree said.
The pilot in Burkina Faso ran about six months from onboarding to active use, and the Burkina Faso teams have no intention of stopping. “The teams were clear that the pilot will conclude, but they want to keep going,” Crabtree said. “They can no longer do without the insights these analyses provide.” The nutrition and economic recovery teams have asked for the same approach to visualization, and the regional measurement advisor is planning to deploy dashboards to other countries in West and Central Africa. “We’ve given a lot of thought to the best approaches and options for adopting and integrating artificial intelligence into our workplace,” Attiey said. “Now that adoption seems to be moving so quickly, we’re focusing on how to better guide its use and ensure its security, reliability, and integrity,” Attiey says.
“The lesson from Burkina Faso is not simply that AI can make analysis faster,” Crabtree added. “It’s that, when paired with local expertise and appropriate safeguards, it can help teams turn existing information into earlier action—strengthening how programs respond to the needs of the people and communities they serve.”
“One of the most exciting parts of this is that it's going to let us turn data we already had but couldn't visualize into something we can actually use to make the decisions that we want to make”