As Jio turns 10, we celebrate an India that has grown up with technology and is now ready to shape what comes next.
The next decade will be about India leading from the front.
A generation for whom the only question that matters is not “Bade hoke kya banoge?” but “Bade hoke kya banaoge?”
Jio will be there to power every idea, every innovation and every dream.
Aage aage dekho, kya kya banayega India.
#10YearsOfJio
A dumb AI model just destroyed 188 football fans at predicting the World Cup.
It took 3 hours to build. It’s only rule: always pick the favourite.
71% accuracy. Beat 92% of humans.
Consistency beats intelligence.
Full breakdown: https://t.co/G9DA423JhJ
A dumb AI model just destroyed 188 football fans at predicting the World Cup.
It took 3 hours to build. It’s only rule: always pick the favourite.
71% accuracy. Beat 92% of humans.
Consistency beats intelligence.
Full breakdown: https://t.co/G9DA423JhJ
You don’t need to know anything about football to predict it better than fans.
We tested this during the World Cup.
A model with zero football knowledge (just Elo ratings from Wikipedia) went 74/104. That’s 71%.
188 football fans who watch every match? Average: 62%.
The model’s only rule: always pick the favourite. Never pick an upset. Never go with your gut.
That one rule beat 92% of humans.
Consistency beats intelligence. Every time.
Full experiment with data: https://t.co/H0oSuh6gVL
The whole thing was built with AI @ManusAI in ~2 weeks. 15,000 lines of code, 502 tests, 7 automated cron jobs, zero manual intervention during the tournament.
Full writeup — model details, behavioural patterns, engagement data, technical architecture:
https://t.co/G9DA423JhJ
188 people at Fynd @fynd_official predicted every World Cup 2026 match.
We also built a model in a few hours using only Elo ratings.
The model finished #1.
Here’s what we learned about humans vs. algorithms 🧵
The model is embarrassingly simple:
1.Elo ratings → win probability
2.Poisson distribution → expected goals
3.Pick the highest probability outcome
No neural nets. No training. No GPT. Just math that’s been around since 1960.
A proud moment for all of us at Fynd as we add a new landmark to our office premises, a symbol of our journey, our culture, and the spirit that brings us together every day.
@farooqadam@sreeramanmg
We just got featured at the @Google Cloud Next '26 Opening Keynote.
Reliance Retail built Jiia, their AI shopping agent, on Fynd Kaily, our agentic commerce engine.