We Just Built the World’s First AI-Powered Live Sports Journalist for Horse Racing
By Shane McCarthy, CEO of Equitas
Yesterday, at Punchestown—on the hottest day of the Festival, in front of 42,138 people—we tested something we’ve been thinking about for a long time.
It was a day of real energy. Constitution Hill returned to Ireland for the first time since his point-to-point days. State Man caused the upset with a scintillating performance with Paul Townend on board. And we ran a live AI system, in the middle of it all, that just changed what’s possible in sports journalism.
We built and tested what we believe is the world’s first AI-powered live sports journalist for Horse Racing. And it worked magically, well… by the end. The goal was to get it to 90 seconds if possible and at the start, it really didn’t look like we’d get there.
We were testing throughout the pre race to get it setup… it was torture, it just kept breaking… we got to the first race, but the system really wasn’t functioning how we needed it to.
It was not writing in our style, our tone… just poetic and dramatic nonsense… it wasn’t following the modular and linear paths we’d created… it wasn’t writing like a racing journalist… then we’d problems uploading the racing programme and pulling the info, it was making up information and really messing up using ChatGPTs Optical Character Recognition (OCR)... lets just say at that stage, at the end of race 1... we weren’t happy…
We had to take a breath, re-think, iterate and then we had to make some adjustments…
Then it clicked in Race 2.
From there:
- The first race test published in 7 minutes
- Then 6 minutes on the second test.
- Then 4 minutes on the third (at this stage, we’d have been happy with 2 minutes).
- Then, finally— it landed… full flow, no issues, brilliant inputs...
That final test was special.
18:55 – the winning photo of State Man was taken in the parade ring
18:56 – the article went live
We had done it.
1 Minute.
It took us 1 Minute from the winning picture being taken in the parade ring after the State Man and Constitution Hill race.
The final article… in all honesty, it needs work… maybe 80% of the way there…
But we are buzzing… the article we can improve, fix, enhance… the writing style, tone we can scale and personalise… we had gotten the testing data we needed from our first test in a high pressure situation.
The article was a great first step and a pathway forward... a real learning...
It included pre-race context, real-time race narrative, and post-race commentary—all voiced in by a human, all rewritten by the system in that journalist’s own tone and style.
Not templated. Not summarised. Just done—professionally, personally, and fast.
The late stragglers were still walking into the parade ring when the article was published.
We knew what we’d built. And when we saw it go live, we just stood there—me and Muireann—buzzing, a bit wired, and honestly, kind of overwhelmed. We looked at each other and said...
“I really think this doesn’t exist anywhere else in the world I said." "It certainly doesn’t exist in Horse Racing Media, said Muir”.
Well… at least it didn’t until yesterday. Tested for the first time ever on the sacred ground that is Punchestown and a magical day with State Man taking the win.
To celebrate that moment with Muireann, and Tom in Punchestown and to give Sarah a call and put her on speaker— we had goosebumps. Everyone smiling. Everyone buzzing. Just laughing and happy… It has taken a lot of time and effort but we knew: this live sports journalist wasn’t just fast… it was ours.
Built by us. In our voice. With a clear vision behind it. And it worked brilliantly.
It was surreal ringing Sarah and celebrating with Muir straight after. A moment I won’t forget for a long time.
What This Actually Is
This is AI infrastructure for publishing. This is another wave of journalism.
It’s a working system—built, tested, and now live—that allows someone… anyone… to speak what they’re witnessing at an event… and within seconds, have that turned into a full-length, brand-accurate article in their own voice.
The tone is human. The structure is sharp. And it’s designed to match the journalist, the brand, or the publication—exactly.
This system is the first real test of what that looks like in action.
It’s been built, broken, rebuilt, re-trained, restructured. We’ve stress-tested the tone, safeguarded the legal layers, obsessed over brand alignment, trained the system to sound like us, and refined it to keep the human firmly in the lead.
We believe one thing above all else:
AI shouldn’t take the voice away. It should amplify it.
That’s what this system does. It doesn’t publish in some flat, AI-polished format.
It is published in your voice, your tone, your writing style.
Why This Matters
This changes the game for event coverage.
- It’s fast.
- It’s personal.
- It’s accessible.
- It gives people the tools to tell stories as they happen—and to do it in a way that sounds like them, not like a bot.
Anyone on your team can become a journalist. They don’t need to write a word. They just need to observe, speak, and send. The system handles the rest.
This is the kind of publishing power that usually takes teams, time, and post-event effort. Now, it’s live for more testing at upcoming shows and events. And it’s just the beginning.
We have four more systems/engines built or under construction to complement our live sports journalist focusing on events, shows and reactive news AI journalists.
Designed to expand what’s possible with media and publishing across the Equine Industry and throughout media around the world. Designed to make journalism more accessible and reactive… live and purposeful, driven by AI by Human Led and really personalised.
These 4 complimentary systems will enter testing over the coming days and weeks. And we’re excited to see how they connect.
Agent to agent. System to system. Brand to voice.
We want to be leaders in this space. Building the infrastructure that will transform media and publishing and enable us to support women worldwide not just in the equine industry, in all industries to amplify their story.
We’re not trying to automate journalism. We’re trying to make it more scalable, more human, and more powerful—with AI doing the execution, and people leading the experience.
We’re proud of what we’ve built. Proud of how it held up under pressure. Happy with and focused on where this is heading next.
We are really building AI driven media and publishing systems with some real purpose at the minute… wonderful work by our team for phase 1 of this build out and a great result in the first major test for any of our AI Systems and Engines.
It’s exciting, it's real, it's Equitas.
Buzzing with this, cracking work on a special day.
Until next time.
Shane
