Built a "Prediction Bench" for 5 of the top frontier models to predict the #WorldCupFootball games.
Simply asked "Can you help me predict the winner of every game at the 2026 World Cup?" with a spreadsheet of stats from games over the last 4 years.
Bench Scoring: 10 for getting the pick right (W/D/L), 5 variable points based on confidence, 5 variable points on getting the scoreline correct.
Track the leaderboard at https://t.co/uf0UxSo5hb
"We're a family of four with a nine-year-old son and a six-year-old daughter. Can you help me figure out which fast food restaurant would be the healthiest one to stop at and what it is that we should probably order?"
Continue to play with this a bit more -- fun little experiment: https://t.co/GNk9BcaYaD
Wrote an article around a few of my favorite things these days — soccer + AI.
Using a new agent framework, we built an "analyst org" to analyze data and answer questions.
First question: Who will win the 2026 World Cup?
Check out the full writeup: https://t.co/c6xlbsvJNE
Wrote an article around a few of my favorite things these days — soccer + AI.
Using a new agent framework, we built an "analyst org" to analyze data and answer questions.
First question: Who will win the 2026 World Cup?
Check out the full writeup: https://t.co/c6xlbsvJNE
We wanted to kick the tires on @flueframework, the new agent framework from the @astrodotbuild team.
So we built the Flue Analyst Org:
- Ten agents
- Six frontier models
- Your dataset
- Your question
It profiles, checks quality, analyzes, critiques, builds charts, and writes an answer.
With the World Cup kicking off in two days, we asked it who is going to win.
Full writeup from Haydn Strauss:
https://t.co/VZ1TvsZ0GK
Try it here:
https://t.co/K6SQQOP7Ax
Note: This is a for-fun test app. Not a product, betting service, or substitute for real analysis.
@TMFanSupport help! We have tickets to the avalanche game lost in cyber space after a transfer earlier today. Both of our accounts say tickets transferred to each other.