For decades, Colombia’s extraordinary biodiversity remained largely inaccessible. Prolonged internal conflict, cartels, jungle guerrillas and other armed groups limited access to vast natural areas, keeping local and international tourism away.
Following the 2016 country’s peace agreement, those landscapes have gradually reopened. What was once off-limits is now drawing increasing numbers of visitors, positioning Colombia as one of the world’s leading ecotourism destinations, and placing the contyry as at the top of global birdwatching destinations. Yet this resurgence has brought new challenges. As tourism has grown, so has its impact on delicate ecosystems, which were unaccesible for visitors before. Increased human presence, noise, artificial lighting and habitat disruption have begun to affect bird populations, contributing to nest abandonment, altered flight patterns, and reduced breeding success in certain areas. In response, researchers from the Behavioural Ecology & Conservation Group, in collaboration with Universidad del Rosario in Colombia, have developed ‘WingSentry’, an AI-powered system designed to monitor and protect bird populations without disturbing their natural habitats.
According to PROColombia, the country’s national agency for tourism, with nearly 20% of the world’s bird species, Colombia ranks first in global bird diversity. Birdwatching alone generates more than 7,000 jobs and over $9 million annually. “The environmental impact of AI has been questioned, so we gave AI a positive purpose: helping tourism
coexist with the very nature people travel to see,” said Juan Diego Pineda, Biologist and Lead Researcher in the project. “Protecting these species is essential,” he added.
WingSentry combines thermal imaging technology with bioacoustic sensors to observe wildlife with zero habitat disturbance. The value of the system lies not only in the data it collects, but in how that data is applied. WingSentry translates complex ecological information into practical guidance for tourism operators, helping them design bio corridors that avoid sensitive nesting areas and reduce human impact. WingSentry was validated through an extensive field-testing program in Colombia. During the pilot phase, approximately 120 hours of synchronized audio and thermal video data were captured. Preliminary site reconnaissance was conducted in 2024 via four distinct expeditions to Cuevas del Peñón (Santander), each consisting of one-week sessions. Subsequently, the final exploratory data acquisition phase was executed in 2025 at both La Danta (Antioquia) and Cuevas del Peñón. This final stage comprised two intensive deployments, six-week long each. All data collection was integrated with an ongoing population estimation study of the Oilbird (Steatornis caripensis), one of the Andes native birds species.
The data gathering results of the field deployments analyzed the severeal hours of thermal and acoustic recordings, and the Al system analyzed over 50 million data points, as of March 2026. The software and machine-learning system was developed using Python, RStudio, and Jupyter Notebook, and deployed by using the public open source platform BehaveAI. The AI system analyzes millions of data points to identify species, maps trajectories, count populations and detects behavioral patterns, key information to guide tourism operators.
The project was first presented locally at the III Colombian Congress of Speleology and at the VIII Colombian Ornithology Congress; and internationally at the Animal Behaviour International Congress in Vienna. Since then, WingSentry secured support from the Animal Behaviour Society and the Max Planck Institute, one of the most important scientific entities worldwide. And the impact reaches everyone: Strengthened tourism as a sustainable alternative to illicit crop economies, bird species receive stronger protection and travelers enjoyed Colombia’s natural beauty more responsibly.
