Good technology, hard questions: Sleip rollout puts NSW trainers on edge

Cover image courtesy of The Image Is Everything

Artificial intelligence gait analysis is no longer a theoretical welfare tool for racing. It is already being used in stables, veterinary departments and international horse movement protocols to help assess how horses move, monitor changes over time and identify asymmetry with a level of objectivity that visual assessment alone cannot always provide.

Racing NSW’s proposed introduction of Sleip from July 1 sits within that broader shift. The app analyses video of a horse in motion and produces objective gait measurements, giving trainers and veterinarians another data point to consider alongside what they see in front of them.

At a trainers’ meeting on Monday, held to further explain the proposed implementation of Sleip, attendees raised questions about cost, responsibility, consultation, data use, device access and the practical workload involved in filming horses for the app.

Several trainers present said the meeting did not resolve their concerns and some walked out.

What Sleip does

Sleip is an AI-powered gait analysis app that uses smartphone video to assess a horse’s movement.

Unlike some older forms of gait analysis, it does not require sensors, markers or specialist hardware. The horse is filmed on a smartphone, either trotting towards and away from the camera or moving on a circle, depending on the required assessment.

The app then tracks fixed points on the horse’s body and analyses stride pattern, stride variation and asymmetry between limbs. The result is a set of objective measurements that can help indicate whether a horse is showing irregularity in its movement.

For trainers and veterinarians, the value is not in replacing what an experienced eye can see, but in adding a measurable reference point to it.

Lameness assessment will always involve professional judgement. Experienced horse people can often tell when something is not quite right, but subtle changes can be difficult to quantify, particularly with horses that have unusual but normal action, horses returning from injury, or horses being monitored across a campaign.

Used well, objective gait analysis can help show whether a horse’s movement is improving, deteriorating or remaining stable. It can support rehabilitation monitoring, compare movement before and after treatment, and provide a record that can be reviewed by the professionals involved in the horse’s care.

How Hong Kong uses it

Dr Amy Kelly, Veterinary Surgeon for the Hong Kong Jockey Club, said Sleip is used in Hong Kong largely for serial monitoring, particularly with horses that do not move well.

In those cases, the app gives veterinarians an objective measurement they can compare with their own visual assessment.

“It is great technology and often the vets here use it at their discretion, particularly for serial monitoring of a horse, or one that has ‘poor action’,” Kelly said.

“You may look at the horse yourself and think, ‘his action has gotten better’, but I like to use the technology to give me objective measurements and see if it agrees with my professional opinion.”

The app is also used by stables in Hong Kong, although Kelly said its use is at the veterinarian’s discretion rather than a blanket mandatory requirement across every horse in the jurisdiction.

Sleip also forms part of Hong Kong’s veterinary process for international horses. International runners are required to send a Sleip video from their home country before travelling to Hong Kong, before being examined in person by a veterinarian on arrival.

The cost is covered by the Hong Kong Jockey Club.

The Hong Kong use case shows the technology can sit within a defined veterinary framework. There is a clear purpose, human review is built into the process, and the cost is covered centrally.

In NSW, trainers say the same clarity has not yet been provided around who pays, who reviews the videos and what happens after the app produces a result.

The need for human review

Dr Kelly said Sleip can be a useful tool, but only when used as part of a broader assessment process.

“If we are going to use any of these data collection tools, they have to be used in conjunction with looking at the horse,” Kelly said.

“But it is not the be all and end all. The results can be inaccurate if a horse tosses its head or there is similar interference. A human vet watching the horse can see that. I don’t make veterinary decisions just based on the app’s feedback, I always look at the horse as well.”

That practical observation goes to one of the most significant unresolved issues in NSW. If the system relies on uploaded video, the quality and integrity of the footage become central to the usefulness of the data. A horse’s movement can be affected by how it is held, how it is trotted, the surface, the angle of the video and whether the horse is behaving normally at the time.

Hong Kong ensures identifying features, including brands, are included in the videos, but Kelly said any system based on uploaded footage still needs appropriate review.

“Without that interaction with the horse, the data is incomplete,” Kelly said.

Kelly said most participants want to do the right thing by the horse, and even from a purely performance-based perspective, unsound horses are of little use to trainers trying to win races.

In New South Wales, there are several practical questions left unanswered. Who will review the uploaded videos? Will a human assess the footage, or only the AI-generated result? If a horse is flagged as potentially lame or asymmetrical, who receives that information and what action follows?

The scale of those questions is considerable. In the 2024/25 season, 10,304 individual horses raced in New South Wales. If each of those horses were uploaded weekly, the system would generate more than 535,000 videos a year before accounting for horses in pre-training, rehabilitation, early education or spelling.

If human review is intended, the workload would be substantial. If it is not, trainers want to understand how Racing NSW intends to use the information.

Scale and timing

The proposed Racing NSW model, as explained to trainers, would require horses to be filmed and uploaded to the app on a regular basis.

One concern raised at Monday’s meeting was the point at which a horse first becomes captured by the requirement. Trainers present said they asked whether the obligation would apply only to horses in full training, horses close to trialling or racing, or all “active horses”, a phrase they said was used during the presentation but not clearly defined.

The answer has practical consequences. A horse in full work at Randwick is not in the same category as a yearling in its first educational preparation, just as a metropolitan stable with multiple staff and strong internet access is not dealing with the same logistics as a small regional operation with fewer resources and limited reception.

Trainers also want to know how the system would apply to horses moving in and out of NSW. If an interstate trainer brings a horse into the state for a short campaign, it is not yet clear whether that horse would fall under the same requirement, or what value a single upload would provide without a longer history for comparison.

There is also the question of device access.

At present, Sleip is only available on Apple devices, creating an immediate hurdle for trainers and staff using Android devices.

Trainers who attended Monday’s meeting said they asked how the technology would be accessed by those without iPhones, or by trainers in areas with limited reception or unreliable internet. They said they were not given a clear answer.

If the requirement is mandatory, trainers say they need to know whether they will be expected to purchase Apple devices, whether shared stable devices will be provided, how uploads will work in areas with poor coverage, and whether there will be any allowance for those unable to meet the technical requirements immediately.

Cost and workload

Cost was another widespread concern. According to Sleip’s website, the service costs US$219 (AU$306) a month for veterinary level access, with an additional US$150 (AU$210) charged per device the app is accessed on.

Racing NSW says it has negotiated a mass-use plan designed to reduce that cost, but trainers at Monday’s meeting said they were left with the impression the fee may still fall to them. Racing NSW Chief Executive Peter V’landys told TTR the details of implementation were still being finalised.

“While there has been some concern about potential cost, it is important to note that Racing NSW has negotiated a mass-use plan that is designed to be cost-effective and practical for the industry,” V’landys said.

If the fee is passed to trainers, owners or individual stables, it becomes another cost in an operating environment already under pressure from labour, feed, transport, insurance and compliance.

Trainers who attended the meeting said an additional concern was the labour involved in correctly filming horses at scale.

One example raised by attendees was Overpass (Vancouver), who was filmed ahead of his trip to Royal Ascot. According to reports from the Baker stable, the process took 10 minutes to capture suitable footage of a quiet, older horse accustomed to regular trot-ups, with their most experienced staff member on the lead.

Whether that time is typical or exceptional, trainers say it reflects the operational issue. Across an entire stable, the time required to film, upload and manage videos becomes part of the cost.

The consultation process

Trainers who attended Monday’s meeting said they had not yet been meaningfully consulted.

Trainer Brett Partelle has written to the New South Wales Trainers’ Association board, in correspondence shared with TTR, saying the technology had been “presented without prior consultation with trainers” and calling for clearer engagement with participants.

“While we acknowledge that innovations may have a place in our industry, the current approach feels imposed upon us without adequate consideration of trainers’ perspectives,” he wrote.

Partelle said the answers given at the seminar had added to concerns rather than resolved them.

“The responses were confusing and highlighted a disconnect between Racing NSW and the NSW Trainers Association, as well as the realities we face on the ground,” he wrote.

He said the Sleip dispute had also become part of a broader frustration among trainers about certainty, consultation and transparency in NSW racing.

“The industry is at a tipping point, with ongoing concerns that have been raised repeatedly over the years,” Partelle wrote. “There is a palpable lack of certainty and leadership in New South Wales racing, exacerbated by a disturbing lack of transparency. Many participants are reaching their breaking point, and it is evident that we need change.”

NSWTA CEO Richard Callander has publicly called for Racing NSW to consider a trial before full implementation. Several trainers who attended Monday’s meeting noted that Callander was not present at that meeting.

The change-management view

Tracey Steele, a change-management consultant who has worked across projects for major organisations, said technology rollouts often run into difficulty when the people expected to use a system are not brought properly through the change.

“The big ticket with change management is about who is impacted, how they are impacted and what those impacts could be,” Steele said.

“Most changes that fail are about lack of awareness and lack of desire. Getting people to use a new system relies on asking the end user, ‘What’s in it for me?’ Why do I need this app? Are you going to use this to punish me? If you don’t answer this, you’ll get resistance to change,” she said.

“It comes back to the why. Why is this being implemented? What problem is being solved by having it? Where does the liability sit?”

For trainers, those questions extend beyond the app itself. If the app flags a horse and the trainer or their veterinarian disagrees, what process applies? If Racing NSW receives the data, does that alter the relationship between regulator, trainer and veterinary practitioner? If a trainer fails to upload a horse, what are the consequences?

Steele said a pilot program can help identify practical issues before a broader rollout, although it needs to test the full range of users rather than only early adopters.

“A pilot program works as a stress test, to work out what works and what doesn’t,” she said.

Part of a broader strategy

Racing NSW says the proposed use of Sleip forms part of its broader safety and welfare strategy.

V’landys said the initiative was linked to this year’s trainer education focus on reducing musculoskeletal injuries.

“Racing NSW was the first jurisdiction in Australia to introduce mandatory professional development for trainers,” he said.

“These initiatives reflect our ongoing commitment to education, early intervention and continual improvement. The proposed use of the Sleip lameness detection app is another step in that direction.”

V’landys said the purpose of the technology was to support best-practice assessment and monitoring, helping identify potential lameness concerns earlier and more consistently.

“Ultimately, this technology is intended to further protect the safety and welfare of both the horse and the jockey,” he said.

“We will do everything within our power to reduce risk and prevent the kind of tragic accident that could cost a jockey their life. Safety must always remain a priority, and Racing NSW will continue to take a proactive approach where technology, education or improved processes can assist.”

He said Racing NSW would continue to consult with the NSW Trainers' Association as part of the implementation process.

“We value that engagement and recognise the importance of ensuring trainers are properly informed, supported and involved as this initiative is introduced,” he said.

Sleip declined to be interviewed about the NSW rollout, saying questions relating to Racing NSW policy and implementation were best directed to Racing NSW while details were still being finalised.

Sleip also confirmed a number of trainers in New South Wales have already been using the technology for some time, although voluntary use is different from a statewide mandate. A trainer choosing to use the app can decide when, how and why it fits into their own system.

A mandatory model requires broader policy settings, including cost, access, training, data ownership, review process, compliance consequences and provisions for stables with different levels of staff, technology and connectivity.

As of this week, trainers are still seeking those answers.