AI for today's challenges

Electronics & electricals need their own AI to combat climate change.



We train on a broad set of unlabelled data to do a wide variety of tasks


We apply information about one situation to another


geospatial, event-sequence, time-series, and other non-language factors

Your organization + climate science + Paris Agreement

How can you determine your emissions and meet your responsibilities?

By March 1994, 198 countries had ratified the United Nations Framework Convention on Climate Change (UNFCCC). Parties to the Convention committed to act in the interests of human safety - even in the face of scientific uncertainty . The stated goal is to limit global average temperature increases to as close to 1.5C as possible above pre-industrial levels. The electronics industry is responsible for potent greenhouse gases.

Countries are able to set straightforward goals. Organizations, however, struggle to define their part in supporting the Paris Agreement. Perhaps you estimate your contribution per country, based on data published by local and federal governments? Or maybe you employ a bottom up approach that relies on your direct and indirect emissions? Either approach will rely on a fair degree of estimation.

That is where AI comes in.

With MobiCycle AI, we use data from operations and from every part of your value chain. Upstream activities such as materials production, preparation and processing are pulled from survey results and remote sensors. We then test these estimates for reasonableness by comparison with satellite earth observations, ground-based measurements and field visits. Finally, we include data published by countries such as policies, measures or an annual inventory of their greenhouse gas emissions.

Our estimates are used to answer the following questions:

  1. Can we predict climate change emissions with AI?
  2. Can we reduce or mitigate climate change emissions with AI?
  3. Does AI's carbon footprint mean its use should be limited?

How do we do it? First, we digest the data. The algorithm gets better with more data we obtain. For example, our predict function iterates 30 thousand times over the data to determine the best prediction.


We bring together data from a range of sources to develop the most helpful picture to guide decision-making.