Thinking Machines Data Science

Helping organizations restore nature and reduce poverty by using geospatial data, AI, and earth observation data.

Past and Current Partners

UNICEF Innovation Fund, UNICEF Venture Fund, Conservation International

Active Countries
Indonesia, Philippines, Colombia, Singapore, Thailand
Thematic area(s)
Climate, Inclusive Growth
Open Source, AI
Organisation Name
Thinking Machines Data Science

The Problem

Conventional data collection methods - like household surveys and on-ground site assessments - are expensive, slow, and limited in scale. Government statistics are often outdated or too coarse to be actionable. As a result, resources and time are wasted on programs poorly matched to their locations.

The Solution

Thinking Machines helps organizations working to restore nature and reduce poverty to more effectively target and monitor their programs by providing them with more novel, up-to-date, comprehensive, and reliable geospatial data derived with AI from the latest earth observation data.

How it works?

  • Step 1: Understanding clients' goals and identifying limitations in their current data. Defining the target variables, geographic coverage, resolution, target time frame, frequency of update, input and training data, and methodology
  • Step 2: Collect, annotate, assess, wrangle, and process the input data including imagery and open datasets
  • Step 3: Develop a model to produce the target data on a pilot area
  • Step 4: Once the model is performing well, it is rolled out to the entire geography to produce final maps
  • Step 5: Produce reports, interactive maps, decks, or scientific research papers using the data
Digital X Solution Thinking Machines Data Science

Bridging the digital divide

Thinking Machines Data Science in mapping socioeconomic data helps to address the data invisibility of vulnerable communities by providing finer resolution data, revealing differences between small areas. They have also done several projects with telecommunications companies to use more up-to-date and detailed socioeconomic maps to identify locations of emerging demand for improved digital connectivity. Their clients in this industry use this information to make decisions about where to improve and expand digital infrastructure, giving more people access to higher-quality internet service.

Impact and highlights

  • Worked with Conservation International to map the top 250,000 most suitable aquaculture hectares (out of 1.3M) to prioritize climate smart aquaculture programmes (Indonesia and Philippines).
  • Worked with Globe Telecom, a leading Philippine telco, to tag and estimate household wealth for every 50x50m area of the entire Philippines, achieving in just 2 weeks what would take a team of 45 people 6-9 years to achieve via ground truthing.
  • Worked with iMMAP, a humanitarian agency in Colombia, to use AI to map over 400 previously unknown informal settlements of Venezuelan migrants.

Plans for expansion

Thinking Machines is working with UNICEF to develop poverty estimation models for nine countries across Southeast Asia including the Philippines, Cambodia, Myanmar, Timor Leste, Malaysia, Thailand, Vietnam, Indonesia, and Laos and AI-generated maps of air quality in Thailand. They are also working with a conservation non-profit to use AI to map aquaculture and mangrove restoration sites in Indonesia and the Philippines.