Our Unique Approach: Human-First, AI-Enhanced
We often get asked if we use AI to label our data. Short answer: no. At Framesports, every single data point is captured by a real person. Our analysts watch your footage and manually tag every event - AI doesn't decide what happened on the pitch. What AI does is take that human-labelled data and turn it into powerful insights, coaching tools, and content. Here's exactly how it works.
Our Process
1. We Train Analysts From Scratch
We recruit and train analysts from around the world through our own online training programme - creating flexible, freelance opportunities for people in communities where remote work can be life-changing. Our analysts work from wherever suits them, on their own schedule, with no commute and no office required. They learn the rules of rugby inside out, then get certified on our custom-built, in-house data entry system. We got tired of every rugby coding tool on the market falling short of what we needed, so we built our own. This lets us capture data to our exact, bespoke specifications - not limited by what off-the-shelf tools allow.
2. We Break Footage Into Tiny Sections
Once an analyst is trained and a client uploads their match footage, we break that footage into small, focused segments. Instead of asking one person to code an entire 80-minute match - the traditional approach - we distribute bite-sized chunks to analysts across the world. This means each analyst can give full attention to a short section, which dramatically improves labelling accuracy.
3. We Quality-Assure Everything
Every event goes through a quality assurance process. Analysts must independently agree on what an event is before it's accepted into the dataset. Because the process is anonymous, there's no groupthink - analysts can't see each other's answers. This consensus-based approach significantly increases the accuracy and reliability of the final data.
4. You Stay In Control
Although we aim for 100% accuracy, we know users sometimes see things differently. That's why we've built in features that let you edit any event we've identified. If you spot something you disagree with or want to reclassify, you can change it instantly. You're always in the driver's seat.
What AI Does
Once our human analysts have labelled every event in your footage, AI takes over on the output side. It compares thousands of historical data points to uncover patterns - why a kicking battle failed, which players consistently dominate collisions, how your scrum compares to averages. Think of us as your extra assistant coach, giving you more time to focus on what matters - your people.
Why Humans Still Matter
Rugby is messy. Weather changes, scrums collapse unpredictably, off-the-ball movement matters enormously. Pure AI misses things more often than people think. Our analysts bring accuracy and context; our AI brings speed and scale. Together, they deliver insight you can trust.
Crafted Alongside The World's Top Rugby Minds
Every feature we've built has been battle-tested by professional analysts. We refine our workflows relentlessly to save time and surface the trends that matter most. Rugby is all we do, and that singular focus is why we're so good at it.
Our Vision
We actually started by trying to go 100% AI from day one. But without sufficient clean, rugby-specific labels, even the smartest models fell short. So we brought humans into the loop, and every match since has added to a growing library of perfectly tagged data. That foundation gives us the best chance of unlocking full automation in the future - but right now, humans are what make Framesports accurate.
