Predict new high value crop strain performance based on past crop trends, weather and soil data. Effectively speeds up traditional farming innovation but can face some PR challenges.

Specify crop growth techniques based on individual plot characteristics and relevant real time data. Scaling out best practice in agriculture will be potentially transformative when applied to the world's nutrition challenges.

Predicting risk of health issues can be complex - especially when the creatures in question are not mammals. One example solution is using an app to listen out for the health of a colony of bees. The system has been trained to recognise sounds indicating stress or disease that may have afflicted a beehive.

Construct detailed map of farm characteristics based on aerial video or image capture. This can be of assistance for a series of tasks ranging from planning to valuation to deployment of automated farming tools.

During the initial phase of drug research and development, the target for a new drug treatment must first be chosen. The target molecule will be what the drug compound interacts with to get the intended outcome, often the treatment of a disease. Researchers use deep neural networks to predict molecular level interactions to treat a condition or improve particular functionality

Increasing production yields through optimising system-long inputs from team, machine, supplier and customer requirements through machine learning.

Optimise pricing (ideally in real time) based on factors such as future market, weather, and predicted crop yields. Clarity on pricing can be of especial assistance to those at the production end of the market chain.

Yield management by taking sensor data on soil quality - common in newer John Deere et al truck models and determining what seed varieties, seed spacing to use etc. Also used for checking the right moment to bring livestock to market - using cameras for example.

Predict real-world results from fewer experiments to reduce experimental R&D costs (e.g., new crop testing). This can have a positive environmental impact.

Deploy robots to do physical tasks in the agricultural process, for example planting, watering and harvesting. Continuing the culture of automating agriculture started in the 18th Century.

Environmental control system for automated greenhouse plant habitats is trained on a dynamic climate model. This then enables it to optimise inputs of energy, moisture and nutrients to maximise plant growth productivity. The need to optimise space utilisation for urban farming helps justify the investment required.

Predict yield for farming or production leveraging IoT sensor and other relevant data. This can work for both plant and animal yield and should also help with planning responses to disappointing predictions,.

Mobile testing of soil and water for chemicals radically speeds up the process and can enable local decisions on critical matters in a time-sensitive fashion. There will be a library of chemical indicators that expands over time. The application might work in conjunction with a physical testing device or product and the AI will interpret the signalled outcome.

Optimise farming production process, often in real time, determining where to dedicate resources to reduce bottlenecks, cycle time and error rates. Interventions might range from automated fertiliser or pesticide deployment to alerting for human involvment.

Using images to identify plant type will enable wider access to up to date agricultural knowledge - whether requiRed remotely (for example to track disease spread) or locally (to support better farming, gardening or rambling).

Predict failure and recommend proactive maintenance for farming and production equipment saving costs and reducing downtime.

Optimise purchasing mix across suppliers and locations should enable better pricing negotiations and reduced wastage on purchased product.

Rather like humans animal health and welfare is strongly related to the exercise and diet that they experience. Hence using data generated by wearables (attachables...) and other sensors to predict animal outcomes should help produce better farming yields (in both quantity and quality terms).

Declines in key populations of bees have triggered multiple attempts to deliver pollination via miniature robots or drones. This remains at a research level - a process which may also be used to learn more about these vital but endangered creatures.

HIgh value animals can need identifying at critical moments - for example to ensure that the right horse has been entered in a race or has been provided to stud. Individual eye scans are kept on record and then a portable scanner can be used to ensure that the right aninal is present.

Machine learning can be used to speed up the product research and development phase of the chemical industry.

Optimise blend mixture and process timing of raw materials being used in refining and similar processes. Sustained small marginal improvements can have a significant cost impact - especially where human involvement can be costly or potentially hazardous.

Optimise mine investment and plans based on drilling samples, past sites, and other data. This may include publicly available data and specially collected private data.