As AI capabilities evolved at breakneck speed, the DevExplorer team rebuilt their tool three times in a single year. This blog reflects on the technical and organisational lessons learned from building an LLM-powered product for international development in a rapidly shifting AI landscape.

Despite growing confidence in AI, there’s more appetite than ever to experiment with and learn about it in safe, low-risk ways. Here are three ways we've been intentionally creating space to play with AI.

While AI tools for monitoring forest cover exist already, these solutions are often trained on data that does not apply across other contexts. Our pilot in Tanzania found that impact depends on local skills, institutions, and long-term ownership.

Can banana-fibre paper reduce nematode-related losses in white yam production in northern Ghana? Our two discovery reports outline the promises of a novel nature-based solution and the practical steps towards real-world adoption.

What happens when climate-driven forest fires outpace the capacity of existing fire detection and response mechanisms? Our pilot in Pakistan looked at how AI can enable early-warning systems for forest fires. Read the findings here.

Tackling the illegal wildlife trade is made harder by the limited tools to accurately identify species and determine their origin. Our MinION pilot in Colombia looked at portable DNA sequencing for better wildlife protection.

Gaps in WASH asset data mean that broken water points often go undetected, disproportionately affecting remote and marginalised communities. This pilot looked at how AI-assisted validation could strengthen oversight and ensure WASH infrastructure issues are addressed before they escalate.

Rapid digitisation of local health systems risks leaving behind those most in need of services. This report examines how practical digital features can improve visibility of vulnerable patients and support more equitable outreach.

Driven by the rise of AI tools and the explosion of the global datasphere, governments are facing a rapid shift in how data is created and used. This blog examines the new skills and organisational changes needed for governments to use AI for data analysis responsibly and effectively.

Concrete and other high-carbon materials dominate construction, driving up emissions and costs. This report explores how bamboo and straw could offer scalable, low-carbon alternatives for sustainable housing in resource-constrained regions.

Health chatbots don’t succeed because they’re clever, but because they’re embedded in trusted systems with clear referrals to real people at the right time. Here’s what we’ve learned over six years of pilots.

Emerging due diligence and traceability requirements across forest commodity supply chains risk excluding smallholder farmers and producers. This report looks at the potential for open data tools to demonstrate compliance.

Budgetary constraints and high case backlog pose significant challenges for Malawi’s legal system. Our discovery report contains research findings on how AI can improve efficiency and enhance access to justice within Malawi.

Read the final report for our pilot ‘Using AI to scale access to forest carbon markets’

Read the report for our pilot 'Smart GeoSeals to track delivery of humanitarian aid' here

Read the report for our pilot 'EmpatIA - enhancing healthcare access to remote areas in Peru' here

The global demand for humanitarian aid is surging, with 308m+ people expected to need assistance by 2024. Traditional supply chains are struggling to keep up, and the pressure mounting to deliver aid faster, smarter, and more efficiently has never been greater.

Explore learnings from our pilot, ‘EmpatIA — AI to enhance healthcare in remote areas of Peru'.