Tomasz -

Covid-19 crisis shows fragility of food supply system

Food system is “incredibly interdependent, but also incredibly disconnected”, says Richard Tiffin, chief scientific officer at early-stage agricultural data company Agrimetrics

Richard Tiffin, chief scientific officer and founder of Reading-based agricultural data firm Agrimetrics, says the Covid-19 public health crisis is revealing the fragility of the global food system.

Tiffin is director of the Centre for Food Security at Reading University, as well as being a founder of Agrimetrics, which was originally funded by Innovate-UK under the government’s Agritech strategy in 2017. The firm’s declared goal is to improve access to, and integrability of, data for use by the agri-food sector.

He draws an analogy between the global food system and the financial system. Both are complex systems inherently vulnerable to small shocks and lacking a “single point of understanding”.

The 2008 financial crash was the catastrophic event that drove Tiffin to form the company of which he is now chief scientific officer.

“Our food system is incredibly interdependent, but ironically it’s also incredibly disconnected,” he says. “We have no idea of the long-term consequences of events like Covid-19. To avoid collapse, we need to understand these connections.”

And so the thesis of the company is more about the resilience of the food system than it is about “feeding the world”, says Tiffin.

“We are today in one of those extreme events,” he says. “Our food system is not really capable of coping. The best example of that is that we have two parallel food systems, one providing the domestic market and one the catered market – restaurants, and so on.

“We have shut down the catered one and are having all sorts of problems getting food to the domestic market. That disconnect could be resolved with a connected data ecosystem that would show where problems are likely to be in the food system and take action to correct the course.”

And, for sure, we have all seen the supermarket shelves denuded of dry pasta in recent weeks.

Data Marketplace

Agrimetrics has built a graph database using GraphDB, dubbed a “Data Marketplace”, which contains global food system information, and which is intended to be used by companies specialising in artificial intelligence (AI) and application development, and focused on agriculture as an industry.

The data sources that feed the Agrimetrics databases – a Microsoft SQL database as well as the graph one – include the UK Met Office, a soil database from a Netherlands research organisation, Nasa, Defra, and satellite data from Airbus.  

Germany-based agrochemical business BASF and Unilever have been early customers. Airbus has also recently said it will use Agrimetrics to sell satellite imagery that can be used to monitor crop health.

“The marketplace is a great way to get people sharing information in the same place, but that’s only the first stage,” says Tiffin. “If we are to answer complex questions, then data needs to be organised in certain ways. This includes predicting future food shortages, finding ways of reducing agriculture’s carbon footprint, and limiting the spread of crop disease.”

Read more about environmental analytics

  • Agricultural AI yields better crops through data analytics.
  • From 2016, Defra minister Rory Stewart plugs data analytics at IEA launch.
  • Reading Institute for Environmental Analytics appoints first CEO.

Agrimetrics’ projects have, the company says, reduced the flow of pesticides into waterways and used predictive models to improve fresh food supply chains and animal health.

Tiffin says it has designed a “Natural Capital Explorer” tool to enable decision-makers to balance out various competing demands on the natural environment from food production, flood prevention, and so on.

It has also developed a “When to Go” app which shows farmers the optimum time to apply an agrichemical to a field. This benefits them because the chemical is not leaching out of the field, but also benefits the local water company because the quality of the water is not harmed.  

But Tiffin has grander ambitions than this. “Average crop yields are around half of what is theoretically possible – we call this the yield gap,” he says. “In theory, if we could better understand why this gap occurs, then we could increase output on arable farms with no additional inputs.

“The best manufacturers have used data and AI to increase production by 50% and cut waste by 20% – agriculture can do the same. It is possible to sustainably feed everyone on our planet for many years to come, but not without AI and not without data-sharing.”

£12m government funding

The company received £12m in government funding in 2017, and now employs about 40 people, including five data scientists conducting semantic data and machine learning modelling, and 10 developer-engineers who work on its cloud infrastructure instance on Microsoft Azure.

Tiffin says the company’s work can be described as “artificial intelligence” because “the semantic integration of data is one aspect of AI”. He adds: “We are also doing this to make it easier for other organisations doing more conventional AI to be able to access the data. I envisage a sea of app developers drawing data out of our infrastructure to create services for farmers in sub-Saharan Africa to multinational companies such as BASF.”

Agrimetrics is also involved in some Microsoft programmes, such as its AI for Good, AI for the Environment and AI for Earth. Tiffin describes the leader of that, Lucas Jopper, chief environmental officer at Microsoft, as “truly visionary”. Under that programme, Agrimetrics is extending its data about fields in the UK to a sample of the world’s fields to produce a global database that will enable analysis of impacts of agriculture on the environment.

The company is also establishing a “data trust” with the Open Data Institute, Tiffin confirms. That will constitute a legal mechanism whereby farmers will be able to share data, under the auspices of a board of trustees, and Agrimetrics will be constrained in what it can do with that data. “They are unwilling data sharers, usually,” he adds.

Read more on Big data analytics

Data Center
Data Management