In my approach to agentic statistics, the agent creates friction on purpose. It costs you a few minutes per step but it is the difference between being in or out of the loop.
wordalisation.com/courses/ag…
My full agentic workflow for statistical analysis on GitHub
Ask Claude to analyse a dataset and it hands back a finished report in one pass. It looks right, and you have no idea which choices were made. This is built to stop that.
github.com/Vivid-Insights/hu…
David Sumpter retweeted
Football analysis doesn’t always happen in one language 🌍
Whether you’re preparing reports for agents or sharing insights with coaches and sporting directors, Earpiece helps you work across languages.
Ask in Spanish and get your answer in Spanish. Or ask in English and request a report in German.
No settings to change. Just ask.
🎧 See it in action.
Available in 8 languages, with more to come!
#FootballAnalytics #AIinFootball #Earpiece #Twelve
A new course, on agentic AI for statistics and data science. How to make an analysis end to end: messy data in, question fixed, figures, tests, and a report someone could review.
wordalisation.com/courses/ag…
AIAIAI is this Tuesday afternoon in Uppsala, and there is still room.
Otema Yirenkyi and Kweyakie Afi Blebo on AI talent in Africa. Victoria Sitati on data science in the age of agents. Mikael Johnsson of Oxx on where software goes next.
doit.medfarm.uu.se/bin/kurt3…
Third speaker at AIAIAI in Uppsala on 20 October: Victoria Sitati.
She coordinates Kujenga, a six-week maths and AI course running on seven African campuses at once — and is asking what you need to know to work in data science in the age of agents.
doit.medfarm.uu.se/bin/kurt3…
I explain context engineering with Fighting Fantasy books.
You pick "fight the dragon" or "run away" and turn to a different page. Function calling is the same: the model picks a route, which decides what it sees.
Monday 12 October, 19:00 UK.
twelve.football/community
Most of our work at Twelve these last two years has gone into the shape of the answer: which numbers matter for the question in front of someone, what sits alongside them, how much to say, where to stop.
Masterclass Monday 12 October, 19:00 UK.
twelve.football/community
Second speaker at AIAIAI in Uppsala on 20 October: Mikael Johnsson, co-founder of Oxx, investing since the dot-com boom.
Most events about AI and Africa have the researchers in the room and the money somewhere else. This one has both.
doit.medfarm.uu.se/bin/kurt3…
You know a log-rank test compares survival between groups. Knowing that has never once produced a p-value.
Generative AI writes you plausible Python and hands it back. An agent installs the package, runs it, reads the error and fixes it.
wordalisation.com/courses/ag…
The people in football with the best ideas about what to build are usually the ones who cannot build it.
Vibe Coding for Football starts 9 November — turning ideas into working apps with AI, for people with no coding background.
Free webinar Wednesday.
twelve.football/webinar/foot…
These tools do not have ideas of their own. They take an idea you already have and formulate it clearly, straight away.
Writing is how I find out whether I understood something. Some of the joy of explaining my own thoughts has gone.
wordalisation.com/courses/ag…
Football analytics still begins with getting a file into the right shape.
Football Python 360 starts 12 October — for coaches, scouts, analysts and journalists rather than programmers.
Free webinar Wednesday on how it all fits together.
twelve.football/webinar/foot…
The richest Premier League clubs will have ten to fifteen data scientists by the middle of next year.
For a decade the bottleneck was not the analysis. It was that nobody above them was listening.
Free webinar Tuesday on how to get into this.
twelve.football/webinar/foot…
On 20 October we are putting on a free afternoon in Uppsala on AI and African innovation.
Otema Yirenkyi and Kweyakie Afi Blebo of Because She Can are speaking. They have reached over 10,000 women and girls in 17 African countries.
doit.medfarm.uu.se/bin/kurt3…
Are Kenyan runners faster than Ethiopian runners? Ten race times from each gives t = −3.52, p = 0.0013.
Then the histogram: the two distributions sit almost on top of each other. Plenty of Ethiopians are quicker than plenty of Kenyans.
wordalisation.com/courses/ku…
From Four Ways of Thinking: the time for the first 3% to see a meme is the same as the time for the other 97% to find out.
Same doubling as a virus. When you hear about something, that is usually the moment everybody else does.
wordalisation.com/courses/ku…
The Vicsek model has fish at constant speed. Real fish burst and glide. Both are used, they contradict each other, and that is fine.
Maths biology got good at analysing models and shy about inventing them.
Drawings by Linnéa Gyllingberg.
doi.org/10.1016/j.mbs.2023.1…
Soccermatics, 2016: scouts sorted a spreadsheet by tackles. Row A was N'Golo Kanté at Caen. Leicester paid £5.6m, won the league, sold him for £32m.
Today the data is a download and the sorting is one click. The work moved to building the columns.
twelve.football/courses
Twelve and Training Ground Guru run a community for people working in football analytics. Six masterclasses this autumn — data engineering, predicting outcomes, tactical formations, context engineering, Amazon, and building a data team.
twelve.football/community


















