About Staibles
Why I built it
Choosing a racehorse share should be the exciting part. Comparing the offers shouldn't be hard work.
Prices, fees and terms vary widely, and every operator describes its offer in its own way. To compare them, a buyer has to read a dozen websites and work out for themselves what a year costs and what's included.
I wanted one place where every offer is read the same way and laid out the same way. Then you can fall in love with the horse, knowing more about what you're signing up to.
How it's built: one person and AI
Staibles is built by one person working with AI. It brings together three kinds of information that rarely sit side by side.
- What operators publish. Websites, FAQs, brochures and terms, each written differently. AI reads them and records every fact with where it came from. Each page is read twice, independently. A fact is kept when both readings agree; when they don't, I read the page myself and decide.
- Racing records. Structured data about the horses themselves, such as registered colours and racing history.
- People. Operators can check their listings before they go live, and tell us at any time if we've got something wrong. Every rule about how offers are compared is a decision I've made, written down and published.
On top of that sits a set of fixed rules that turn facts into a comparison. The same facts always give the same result, and anyone can check how.
Everything else was built the same way: the idea, the comparison method, the data pipeline, the website, the analytics and the automatic checks that stop a wrong fact going live. It took about seven weeks of evenings and weekends, vibe coding with Anthropic's Claude and Claude Code, and just as much time earning the trust of the syndicates and clubs it lists.
What AI does, and what it doesn't
AI does the reading, so one person can cover far more of the market. Fixed, published rules do the comparing.
No AI decides how an offer compares. It doesn't pick favourites, decide what counts as good value or fill gaps with guesses. Where we don't know something, the listing says so.
AI also gets things wrong, which is why every fact carries its source, every operator sees their listing first, and every correction is applied.
Working with operators
The data is only half the job. The other half is relationships.
Every operator is treated the same way. Before a listing goes live, the operator gets a private preview and the chance to put anything right. Several have, line by line, and the site is more accurate for it.
Some of the most useful lessons came from operators pushing back. One showed us that when a yard also sells horses outright, leases them or runs a club, its general owners' FAQ doesn't always apply to its syndicate. That changed how every listing is read.
Who's behind it
I'm Stuart O'Hara. I've had a long career in technology, part of it in racing and betting.
I hold an MBA from the University of Manchester. Staibles is run by my own company, Omnia Platforms Ltd.
Get in touch
Staibles is one example of what a single person can now build with AI: reading messy information at scale, combining it with structured data and human judgement, and turning it into something people can trust.
If you have a similar requirement and need my help, please get in touch at info@omniaplatforms.com.