After Colombia's earthquake, civilians used AI to mobilize aid

After a 7.4-magnitude earthquake struck Colombia in August 2026, volunteers used artificial intelligence to map donation sites and match aid with need when government response lagged. Developer Felipe Restrepo Lebrun built alluda.online with Anthropic's Claude Code Max in two hours, listing 100 relief points across 25 cities within three days. The civilian-led platform emerged as more than 1.5 million people in the region faced outages of water, electricity, and cellphone service.

The quake killed more than 300 people and injured over 4,500, according to officials, with an epicenter in a remote, impoverished area. As donors opened collection points in homes, cafes, churches, and breweries, coordination grew chaotic—some sites had food but no water, others had supplies but no transport.

Key Takeaways

• A 7.4-magnitude earthquake in Colombia last month killed more than 300 people and damaged over 198,000 homes.
• Developer Felipe Restrepo Lebrun built alluda.online with AI in two hours, mapping 100 donation points across 25 cities within three days.
• The site lets users list supplies, request aid, and offer transport—but limited verification raises fraud concerns.
• Regional disaster officials in Risaralda see promise in the tool and plan talks with its creators.
• Similar civilian tech efforts have emerged after earthquakes in Venezuela and elsewhere.

What happened when the earthquake struck Colombia?

The August quake struck a remote, poverty-stricken region where basic services collapsed for more than 1.5 million people. With no water, electricity, or reliable cellphone signal, relief coordination was haphazard from the start.

Civilians opened donation points in homes, cafes, universities, churches, gyms, and even a brewery. Sebastián Mejía, who set up three collection sites at his coffee shops in Pereira near the epicenter, described the effort as disorganized: some locations had food but no water, others had water but no clothes.

Colombia's National Unit for Disaster Risk Management reported that more than 198,000 houses were affected, including over 37,000 requiring total reconstruction.

How did AI help organize earthquake aid?

Mejía messaged Restrepo Lebrun about building an app. Spotty internet delayed the request, but within two hours of reading it, Restrepo Lebrun used two paid Claude Code Max accounts to build a website mapping donation points.

He partnered with brothers Miguel and Juan Camilo Sierra, who promoted the tool on social media. Three days after the quake, users had logged 100 donation points across 25 cities. The site briefly crashed after 23,000 visits.

Originally called Ayudas Pereira, the platform was renamed alluda.online as demand spread beyond the city. Drop-off points can list available supplies, and users can request or offer free goods and services, including transport by car.

Blanca Salazar, who lost her home, learned about hubs through social media. With her granddaughter's help, she used the app to request a tent, bedding, and food for her community in Villa Santana, a poor Pereira neighborhood that went days without power.

Can crowd-built apps replace official disaster response?

Not entirely—but they can move faster when bureaucracy lags. Volunteer Daniel Zuluaga at the Bajo Cuerda brewery said civilian sites were doing work he felt government channels handled too slowly.

Pereira city officials urged residents to rely on a single official collection center to avoid fraud. At the regional level, Diana Ramírez, disaster management coordinator for Risaralda Department, said manual relief work could benefit from technology and planned to meet Restrepo Lebrun with specialist staff.

The project mirrors civilian-led tech responses after earthquakes in Venezuela. For more on how AI is reshaping emergency tooling, see our Future Tech & AI Wonders coverage.

What are the limits of AI-powered relief tools?

Speed came with trade-offs. Electricity and internet remained offline for many survivors, limiting access. With few staff to verify listings, operators have limited ability to stop fraud or exploitation.

Restrepo Lebrun, 44, who volunteered as a teenage firefighter after Colombia's 1999 earthquake, said he could not help on the ground this time because of severe ankle arthritis. He contributed instead through code—and achieved the immediate goal of getting a working tool online quickly.