Navigating the labyrinth of complex data (April 26)

An AI tool is adding value to robotic milking data to support efficient and effective herd management. So how does it work and what does it mean for producers?

TEXT PAUL JENNINGS

Modern dairy units quietly gather an extraordinary amount of information and none more so than herds that are milked using robotic systems. Every time a cow walks into a robot stall, it generates a wealth of information – from milking speed and attachment times to visit behaviour and milk-flow curves.

So for many producers the challenge is no longer about collecting data but knowing what to do with it. It can be difficult to understand which numbers matter most, and how can they be translated into practical herd management changes.

Turning that onslaught of information into something usable can be time-consuming and, at times, overwhelming. But artificial intelligence (AI) tools are rapidly evolving and beginning to provide solutions to these problems.

One such system, developed by Lely, is designed to sift through millions of data points and translate them into clear and actionable advice to improve milking efficiency.

Detailed insights

The AI tool, Field Assist, has already been available in the US and Canada for two years and was introduced in the UK in November 2025.

It is not directly accessible to producers, it is being used by Lely’s Farm Management Support (FMS) team to gain detailed insights into customers’ milking operations. They can then guide producers in implementing the improvements suggested by the system.

Lely Atlantic’s Mike Steele has been training teams across the UK to use the system and says the goal is to harness the knowledge of robotic advisers, along with AI, to achieve ‘gentle’, complete, and fast milking for every teat at every milking for every cow.

Field Assist is an AI tool that learns from feedback, so if FMS advisers disagree with its initial advice after a farm visit, their input helps the system to refine its future recommendations, according to Dr Steele.

The tool integrates data from multiple sources, including Lely’s farm management software, Horizon, and Lely’s internal data systems. It evaluates data at both herd and individual-cow levels.

Performance indicators

It examines a wide range of performance indicators, including: milking profiles; visit behaviour to the robots, connection efficiency, and pre-treatment times; pulsation type; milk flow; and milking speed. The AI tool then identifies patterns and suggests how improvements could be made.

Lely tech offers circular pulsation as an option. Circular pulsation can deliver a more stable vacuum and consistent milk flow than traditional left-to-right pulsation. The AI tool monitors this and can recommend improvements.

“It also identifies slow-end milking by plotting milk flow on graphs and detecting whether milk flow ends promptly or stops and restarts over an extended period,” says Dr Steele. “Slow-end milking can cause overmilking or undesirable pressure on teats, which can extend milking times and lead to teat discomfort,” he adds.

It can be caused by incorrect take-off settings or inadequate teat stimulation before milking. “The AI tool looks at every milking and can home in on individual cows or groups of animals that require attention.”

Milk-access tables

Field Assist can also help to optimise milking robot access. Dr Steele says there is a misconception that the more times a cow visits a robot, the more milk she will produce. But overmilking can become an issue, and if cows visit the box unnecessarily, it limits free time and impacts robot efficiency.

“We are learning from the AI tool that if a cow goes to the robot too frequently – every four or five hours – that this creates a poor milk profile and there is a risk of over milking,” he explains. “It can make milking uncomfortable for that cow and she will give less milk.”

He adds that some of the company’s centres are now delivering real improvements for customers by carefully reducing the maximum number of visits. “On these units, free time has increased, there are fewer milking failures and failed connection attempts, and more milk is produced from fewer robot visits.”

The AI tool also monitors connection attempts, time to attach and connection failures. Then it will identify potential solutions, such as singeing hairy udders, to make it easier for the laser to detect teats, cleaning the sensors more regularly or creating longer milk intervals to allow udders to fill with milk and teats to move further apart.

One of the most valuable features of the data analysis tool is its ability to quantify the potential impact of implementing the recommended changes. “It will work out, for example, that optimising attachment times could gain an addition one hour and 49 minutes of additional robot capacity per day,” says Dr Steele.

The AI tool categorises recommendations by priority, such as high or low, to allow FMS teams and producers to focus on the changes that will deliver the greatest benefit first.

Future role

While the role of AI is rapidly advancing and may still seem overwhelming and a little frightening, Dr Steele believes it offers endless opportunities to automate day-to-day dairying tasks, and producers should embrace it.

“AI automates the labyrinth of complex data and can free up producers’ time to focus on managing cows rather than spending hours analysing spreadsheets.”

That said, he stresses that AI tools are designed to complement human expertise rather than replace it and assist decision-making by analysing data faster and identifying problems before they arise.

Currently, FMS teams are using this tool and others to offer two annual performance reviews with customers to identify areas for improvement.

“It’s another tool in the box, but we still need specialist robot advisers to work with producers to verify the information and implement and monitor changes,” adds Dr Steele.