Fabrizio Serafini

Investment Partner @a16z investing in early stage AI. @Harvard Stats.

Italian export🇮🇹 DJ🪩 Obsessed with sound systems🔊

[Fabrizio Serafini @fabserafini]

https://t.co/spLa3u4AD7

[harsharamesh @harshavramesh]

America's biggest advantage in the race for critical materials isn't a mine. It's our financial markets.

This week @PillarHQ, SIG, and @Kalshi completed the first exchange-traded tungsten hedge for an Ohio metals distributor.

Tungsten goes into munitions, jet engines, and the carbide cutting tools on every factory floor. China produces about 80% of it and exports only to 15 companies. Pentagon rules bar Chinese tungsten from key military uses from 2027. APT prices are up more than 500% since the controls began.

American manufacturers can't wait for new mines to protect their margins. With Pillar, they don't have to.

Bloomberg has the full story. Link below.

[Miguel Ríos Berríos @MiguelriosEN]

Day two of open-sourcing agentrun as @pidotdev extensions: meet pi-browser. Give your agent a browser with persistent context. We use pi-browser plus pi-vision to do complex website analysis. Powered by @browserbase.

[Fabrizio Serafini @fabserafini]

Love this!!! @grepdotai super smart "right model for the right task"

[Miguel Ríos Berríos @MiguelriosEN]

A few weeks ago, we built a new harness, AgentRun, which turns repetitive steps into workflows and code using Jev. This week we're open-sourcing its best features as @pidotdev extensions!

Day 1: Pi-vision uses Jev to give your agents eyes without wasting frontier tokens.

Analyzing videos, images, or documents (a common use case @grepdotai) with frontier models is extremely token-inefficient. With this extension, @typesafeai's Jev does the analysis, while your Pi agent does the orchestration.

The result is a faster and cheaper workflow for detection and analysis at scale!

You can use this workflow for document verification, understanding image content, and needle-in-a-haystack search in videos.

[Fabrizio Serafini @fabserafini]

@AdamHoltererer @ty_geri

[Fabrizio Serafini @fabserafini]

@jgreze One Townie two Townies three Townies...

[Fabrizio Serafini @fabserafini]

I love my Townies.

Huge congrats to @jgreze and the entire @TownAI team!

[Jean-Denis Greze 💡 @jgreze]

The biggest bottleneck at work is waiting on each other.

Today Town is announcing its $108M Series B, co-led by Forerunner Ventures and Menlo Ventures, to fix it with multiplayer AI assistants for work.

Most AI assistants are built for one person. Work doesn't happen alone. I think the moat is multiplayer. Everyone at a company gets a Townie. Townies ask each other questions, share only what their person would share, and do the work. The more a team works together, the more useful Townies get.

What's the thing you're always waiting on a teammate for?

[Fabrizio Serafini @fabserafini]

@kenanhsaleh Forgot to mention Kenan is the 🐐 of @speedrun

[Fabrizio Serafini @fabserafini]

@Chen 10000%

[Fabrizio Serafini @fabserafini]

We spent decades putting business processes into software. But what connects those systems is still a person with 20 browser tabs open, copying and pasting.

A human API.

My request for startups: computer use agents that actually finish the job. Agents that companies can trust to do real work at scale, and that work in the software companies already use.

If you're building this, we want to hear from you!

More below.

cc @a16z @speedrun

[Fabrizio Serafini @fabserafini]

Great great great read from @dhaber

TLDR you can now have persistent AI agents that follow through with customers for weeks and get paid for the results

“Expand the market addressable by AI from a share of labor budgets to a share of revenue”

[David Haber @dhaber]

https://t.co/13UUbuSap0

[Fabrizio Serafini @fabserafini]

@angelajiazhang @jialim__ Doppio Zero!!

[Fabrizio Serafini @fabserafini]

Jev! Jev! Jev!

[Diogo Almeida @CompleteSkeptic]

launched a model

accidentally served trillions of tokens a day

29.4% of the fortune 500 showed up

raised a really big series A from @a16z

it's been 3 weeks

we would like to sleep now

[Fabrizio Serafini @fabserafini]

@mandrusko1 suboptimal

[Fabrizio Serafini @fabserafini]

@seema_amble see below for proof I am an authentic Italian 🇮🇹😂

[Justine Moore @venturetwins]

Someone discovered a new Amazon AI feature that summarizes what it knows about you and I am crying of laughter

[Fabrizio Serafini @fabserafini]

Hot take: very soon most people will stop caring about tokens.

1. Same intelligence gets ~47% cheaper every quarter

2. You don’t care how many CPU cycles an app uses. You won’t care how many tokens it uses either.

What matters is what you do with the intelligence.

[Fabrizio Serafini @fabserafini]

Not only, but remarkable how cheap AI is becoming in the first place!!! Love a good deal :)

[Justin Schroeder @jpschroeder]

No talks about how cheap open source AI models are anymore.

@OpenAI has entirely crushed that conversation with Luna.

[Fabrizio Serafini @fabserafini]

This is an incredible read. AI getting cheaper = you can build more. And more. And more.

[Epoch AI @EpochAIResearch]

AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023.

That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.

[Fabrizio Serafini @fabserafini]

@ShrutiVenkat17 is the 🐐

[Cognition @cognition]

Cognition Ranch 2026.

Three days of ranch life in the California countryside. Here’s a look behind the scenes. 🤠

Come line dance with us: https://t.co/QD5Zf2SOYL

[Fabrizio Serafini @fabserafini]

We are so back 😉

[Olivia Moore @omooretweets]

🚨 The @a16z Consumer AI Top 100 is back - but this time, it's 150!

We added revenue data from @yipitdata to track how consumers are spending not just time, but money, on AI.

The result? A story of heavy concentration among AI's power users. Our takeaways 👇

[Fabrizio Serafini @fabserafini]

🚨The cost of intelligence is collapsing🚨

May: Opus 4.8 Max - AA score 42, $4.08/task.

Sep: GPT 6.1 Sol Med - AA score 48, $0.21/task.

From frontier to outscored at ~20× lower $$. 20x. In four months 🤯

So bullish Apps. Model bill collapsing. Go build something people want.