Nazz🥷🏾

I’m just here to procrastinate 🇺🇬.

[Jurrien Timber @JurrienTimber]

How good is god man

[Bukayo Saka @BukayoSaka87]

You deserve more Arsenal fans 💔

[Arsenal @Arsenal]

The Arsenal. Your Premier League champions.

[Arsenal @Arsenal]

Arsenal: 2025/26 Premier League champions.

[V @TheRetroVay]

Made rival fans hatewatch for 98 mins just to see us win 1-0 😭😭😭😭

[C. @ArsecaI]

Gyokeres against Atletico Madrid's defenders

[Aakash Gupta @aakashgupta]

Google just open-sourced the infrastructure layer for a voice-first economy serving 1.1 billion mobile money accounts.

Africa processed $1 trillion in mobile money transactions in 2024 alone. 81 billion transactions. Over half the world’s mobile wallets live on the continent. But Sub-Saharan Africa’s adult literacy rate averages 66%. Around 30% of mobile money users still need help navigating text-based menus. Every one of those assisted transactions adds cost and fraud risk.

Voice AI that actually works in Yoruba, Swahili, and Hausa removes the literacy bottleneck from a trillion-dollar financial system. The person who can’t read a USSD menu but can speak Luganda gets the same access as a banked professional in Nairobi. That’s a market unlock worth paying attention to.

11,000 hours of speech data across 21 languages, built over three years with African universities and community organizations who retain ownership of what they collected. Makerere University in Uganda collected data for 9 languages. University of Ghana handled 8. Digital Umuganda in Rwanda led 5 more. Google funded the effort but the data belongs to the institutions.

This tells you everything about how the next AI infrastructure race gets won in emerging markets. You don’t extract the data and ship it to Mountain View. You fund local institutions, let them own the output, release it under CC-BY-4.0, and build your models on top of an ecosystem that has no incentive to switch to a competitor.

Microsoft noticed. They just released Paza, a benchmarking tool for 39 African languages. The scramble for Africa’s linguistic data layer is on.

The constraint nobody’s pricing in? 21 languages covering 100 million speakers sounds massive. Africa has 2,000+ spoken languages. WAXAL covers roughly 1% of that linguistic diversity. The remaining 99% still has no usable speech data. The languages left out tend to be smaller, more rural, and harder to collect, which means the cost per language-hour goes up exponentially from here.

Google built the on-ramp. The trillion-dollar question is whether anyone builds the rest of the road.

[Arsenal @Arsenal]

A corner of Arsenal, wherever you are ⚡

Download the new digital home for Gooners now📱

[Adam Karpiak @Adam_Karpiak]

Great news! You’ve passed round 6! Next is an unpaid project and presentation to solve a problem we’re currently having.

[･:*:★ミラ @333cazorla]

british men wake up and the first person they think about is arteta

[JT @jiratickets]

25 year olds with two degrees when the japan trip return flight lands and they're finally out of excuses to postpone getting a job

[LiveScore @livescore]

The Ballon d’Or debate starts NOW — who’s really leading the pack? 🏆🔭

[FIFA World Cup @FIFAWorldCup]

🇺🇬 Uganda's first-ever goal at a World Cup!

#U17WC

[Milwaukee Bucks @Bucks]

🫱🫲

[Next Generation Arsenal @scoutingindoors]

Fuck it

If I see Arsenal conceding a shot on target in the next game Arteta simply has to go

Enough is Enough

Tonight was unacceptable

[Paddy Power @paddypower]

The only 3-4-3 that works in Manchester... 🚌

18+ GambleAware

[Doc @karthikadhaigal]

Eze on Instagram ❤️

[Ballon d'Or @ballondor]

Nominated for the 2025 Kopa Trophy ⤵️

@lewisskelly49
@Arsenal
@England

#ballondor