I was involved in neuroscience research in the 1980s, specifically in the early days of neural network modelling which now underpins most of what we call AI. I took the ideas into sustainability (if you are interested in the details, please see this talk I gave recently at the Royal Geographical Society) and I have retained a keen interest in how we process data, how AI has emerged and how we can’t separate the way we understand the world from the way our brains work. 

Although we have been thinking deeply about how to leverage AI for a number of years, our implementation journey began only in July this year when we launched stage one, our AI Uploader.  This uses an LLM (Claude Sonnet) to enable our users to take any policy, strategy, plan or project in PDF form, convert it into the OnePlanet structure of Outcomes, Actions and Indicators, and automatically create a OnePlanet mindmap.  This functionality now means our users don’t need to do this manually, reducing say a 2–6 hour process into a few clicks and 2-10 min wait. 

The AI Uploader not only saves time. Once in OnePlanet’s database, the data is now structured elegantly, making it much easier to analyse with AI reducing any AI demand to ask questions of the data.  A test shows that we’ll reduce AI demand for querying single plans by approximately 95% and for multiple plans by 99.9%, or frankly even more.  For things like national government strategies or policies, we’ll create a central database so this conversion into OAIs only needs to be done once for everyone in the UK (or whichever country you are in). 

Over the past month we have completed stage two, namely creating our open-source Model Context Protocol (MCP) which allows our users to connect their AI agents securely to their data in the OnePlanet database.  We have also released our own AI agent which you can request to be activated in your account, to make searching your plans and adding data (such as stories and monitoring data) easier and more flexible.  Join us for the Systems Connector Meet Up on 24th September if you want to see how to connect your agent to your data in OnePlanet. 

Over the next month we’ll be working on ‘vector embeddings’ which will, in effect, make every Outcome, Action and Indicator in OnePlanet intelligent – so you can gain insights such how to improve your plans, focus resources and who to collaborate with.   

Being mission-driven, we are very conscious of the negative social and environmental impacts of AI and data centres. Our approach with structured, interconnectable data, promises to help our users solve complex real-world problems incredibly efficiently – and join the dots for joined-up solutions to health, climate, jobs and nature recovery.  This is what we think AI should be used for, not selling us more stuff we don’t need and which is destroying our beautiful planet. 

We have a lot more exciting things in the pipeline.  This includes work we have done with one of the world’s leading neuroscientists and AI expert, Professor Karl Friston, and the launch our cutting-edge AI-enabled functionality to provide you with way to assess the likelihood of any plan or group of plans delivering positive outcomes on health of people + economy + community + planet.  We’ll share more about this and other developments in future blogs and events. 

Watch this space by following us on Linkedin, signing up to our Reimagine newsletter and/or the launch of our ideas for The Great Collaboration.