AI is giving teams a competitive advantage in modern-day rugby

How are the Springboks going to reach the next level in 2026?

Rassie Erasmus will have put this question to his coaches and players in the lead-up to the new international season, which kicks off with a double-header in Gqeberha on Saturday, 20 June.

Several uncapped players are set to feature against the Barbarians and Zimbabwe, and if all goes to plan, the squad will tick the boxes of results and development ahead of the Nations Championship fixtures against England, Scotland and Wales next month.

The quality of South Africa’s performances – and the quality of the information relayed from the coaches’ box to the field – will also be in the spotlight. Erasmus has recruited one of the world’s sharpest analysts in Joe Lewis, and with more information and insight at hand, the team is well placed to progress in 2026.

The point was recently made by Japan assistant coach Gary Gold, in a detailed article about the use of AI in data analysis and coaching at the highest level. Gold has coached various Test and club teams over the past two decades and has witnessed first-hand how data analysis and technology can influence coaching methods as well as team performance.

Having worked with Erasmus at the Stormers (2008) and Boks (2011), he is well aware of the double World Cup-winner’s penchant for innovation – and believes that the recruitment of Lewis could be another masterstroke.

Lewis was appointed as the Boks’ new performance analyst this past March, after concluding his contract with England at the end of the 2026 Six Nations. The Welshman holds a master’s degree in performance analysis and was England’s leading analyst for nine years.

“Knowing Rassie, he will look to take things further in 2026,” Gold told Daily Maverick. “Most, if not all, teams are looking to harness the latest weapon in AI, and Rassie will know how it can give coaches the means to improve their players. Joe will bring his expertise in analysis to the party, as well as his understanding of AI. He will upskill the rest of the analysts while he is there, and will challenge the team to look at the data in different ways.”

Possibilities and pitfalls of AI

In preparation for the Test season, Gold has used AI products such as Claude Code, Grok and Gemini in his analysis of Japan’s players as well as the team’s next set of opponents. Japan won’t play the Boks in 2026, but they will face Italy, Ireland and France in the first phase of the Nations Championship this July.

Gold explains how AI can be a useful coaching and analysis tool in experienced hands, albeit at a later stage of the data-gathering process. “AI allows you to take your trend analysis to another level,” he said. “Previously, you wouldn’t have been able to do that, because you wouldn’t have had the time to go that deep into the numbers.”

He is quick to point out that human analysts and coaches remain vital. “You can’t base your analysis on unverified facts, and it’s for this reason why coding is done by human analysts. They will watch a game and decide what a carry is, what a dominant tackle is, and so on.

“At the moment, AI can’t code live rugby footage due to the complexities around the tackle and set pieces. Furthermore, if you had to give the data-gathering task to AI, some tools might cast the net too wide, and you may get a distorted view of the facts.

“Some of that information might be pulled from third-party data companies, who have their own analysts and standards, which may not be in line with that of a top-flight rugby team. This is an example of how AI can amplify bad or unreliable data.”

It’s in the next step of the process, however, where AI can be a game changer. “The analysts give you the right ingredients, but you still have to decide how you are going to bake your cake – and how you bake your cake will determine it flavour and quality. In simple terms, you need to decide what information to use and how to use it.”

After the conclusion of the 2025-26 Japan Rugby League One season, a team of analysts provided Gold and other national coaches with a comprehensive set of stats – about five million data points collected over 20 rounds of the tournament.

“Once I had all that information, I wanted to go a bit deeper,” he said. “What was really interesting to see, and what AI revealed, is how the existing data can be used in different ways.

“As a defence coach, I might want to know how many post-contact metres one of our Japanese players has conceded, and how many they have conceded against a big loose forward such as Jasper Wiese or against an elusive runner such as Cheslin Kolbe, and in which part of the field. This is important information to have when you are assessing a defender’s stopping power in specific situations.

“Now, human analysts will code how metres have been made by an attacking player and how many tackles have been made by a defender at various points on the field and at different stages of a match. They won’t code how many metres have been conceded by a defender, although using the existing data, they would be able to work it out – although it would take a long time to do so.

“Instead of doing that, I can ask AI to invert the ball-carrying stats – drawing on the raw data that’s already been coded – and the AI will provide me with the information I’m after. I might ask how many metres our player has conceded against a player like Jesse Kriel, for example... Two minutes later, AI will give me what could prove to be a crucial piece of information – and that’s when you start to realise how powerful this tool actually is.”

Boosting the matchday process

The impact that this technology has had on matchday has been particularly profound.

“It’s important to stay focused on matchday, and to ensure that the level of information coming into the coaches box is accurate,” said Gold. “The players will demand that, too, during those brief water breaks and in the 15-minute break at half-time. There could be one or two new things that are coded live, such as a new set up at a lineout that hasn’t been seen before. That will be flagged by the AI and relayed to the coaches quickly – and from there, the group will have the chance to adapt.”

In response to Gold’s article, some have suggested that rugby’s reliance on data analysis and technology will render the leading teams more predictable and one-dimensional. And yet, it’s plain to see how more information – delivered in a quick time thanks to AI – will give teams more options, and the means to respond when things don’t go their way.

Gold believes that these tools have pros and cons and should be handled with care.

“There’s so much potential here, but there are also red flags. AI amplifies the raw data, so if the first step in the collection process is flawed, the AI used in the next step may give you the wrong picture. You also have to accept that AI will give you more information rather than less. You could end up doing twice as much work if you’re not careful, or focusing on the wrong thing altogether.

“Coaches and analysts are always going to make key decisions that shape the way the team plays – and again, there’s always going to be an emotional side of the game that goes beyond the numbers. AI may change the approach to data analysis, but it’s never going to replace the human element completely.” DM

This story first appeared in our weekly DM168 newspaper, available countrywide for R35.