Alysa Liu recently joined the @a16z new media team to help shape the narrative arc of the future. "I’ve spent enough time on the internet to know it’s the most important place in the world. It’s not separate from real life anymore. It is real life". She adds, "we’re building the team that starts, shapes and shares the most important conversations—for our founders, but also for tech."

We’ve published a new blog with @googledevs on how to build a smart financial assistant using LlamaParse, our state-of-the-art agentic document parser, together with Gemini 3. In the guide, we show how to: 📝 Parse a financial PDF with LlamaParse 📊 Use VLM-enabled agentic OCR to accurately extract text and tables ⚡ Combine the parsed content with Gemini 3 to generate a clear, human-friendly financial report 📚 Read the blog: https://t.co/Bw1GRpuBXm 👩‍💻 Explore the repo: https://t.co/HLtlxaf23T 🦙 Get started with LlamaCloud: https://t.co/wZjhFV29gN

ACRNM ◤◤◤ J104-WS white → absolute grail https://t.co/GwJZeiFMzD

ACRNM ◤◤◤ J104-WS white → absolute grail https://t.co/GwJZeiFMzD

🚨NEWS: SENATE MAJORITY LEADER JOHN THUNE spoke to President DONALD TRUMP Sunday and offered to fund DHS -- without ICE. THUNE said he'd fund ICE in a reconciliation package. TRUMP rejected the offer. Said he wants Rs fighting Dems over DHS funding, SAVE Act. Some Senate Rs and WH beginning to diverge as DHS shutdown nears 40 days. https://t.co/Ap81c9nisf @bresreports @LauraEWeiss16 @sn_handler and me

🚨 BREAKING: Meta researchers showed a model 2 million hours of video. No labels. No physics textbook. No supervision at all. It learned gravity. Object permanence. Inertia. And it just beat Gemini 1.5 Pro and GPT-4 level models at physics understanding. Here's what just happened:

ICE agents run from the press inside JFK airport as they are deployed with no masks. https://t.co/fkYLcKa0fU

ICE agents run from the press inside JFK airport as they are deployed with no masks. https://t.co/fkYLcKa0fU

@Babygravy9 I really like this edit https://t.co/7UzbQ1RyhO

Let your AI agent read any PDF on the internet in seconds 🌐⚡️ ``` curl -sL https://t.co/dt61c4zQsi | lit parse - ``` LiteParse is our fast and free document parser designed to seamlessly plug into 40+ different agents. Includes both text parsing and screenshotting capabilities; it doesn't use any VLM under the hood so you could run it on any device. Our latest update allows liteparse to support URL parsing and buffers/streams 🙌 Repo: https://t.co/JNER0mVcB8 Guide: https://t.co/eAj5lhEReM

+1 more

I asked claude to estimate the prices of things based on his data from 2023/2024 and then check the actual price, and show me a graph of the difference 3% inflation my ass. 70% in 2 years https://t.co/FbFHgyOsql

What did humanity lose that stopped us making things like this? https://t.co/Po60Kguh8w

What did humanity lose that stopped us making things like this? https://t.co/Po60Kguh8w

The new Deep Think is very, very smart, but is constrained by the Gemini interface compared to GPT-5.2 Pro. If I can't see actual work in the thinking trace, or get downloadable files as proof of work, or see evidence of the code & statistics it applied, checking results is hard https://t.co/Gyhz3SMVs9

You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only. https://t.co/sVymgmtEMI

2026 prediction: More code will be written this year than in **every prior year of human history combined** 🔥 The vibe coding explosion is real—thanks to these tools turning ideas into apps instantly: - Replit → https://t.co/4ilG1qEld4 - Cursor → https://t.co/9C9t2tKNg1 - Lovable → https://t.co/ekQ7L0oIh4 - v0 by Vercel → https://t.co/DF09AdCgli - Claude Code → https://t.co/QZVBumbnW6 (or https://t.co/uCbCZMuy7d) Prompts → production code. Who's shipping what? 🚀 #VibeCoding #AI

+3 more

Growing my hair out starting today https://t.co/quxhadRLqC

Quickly vibe coded 3D breakout game with GitHub Copilot to use face tracking to control the paddle (and also adjust the board so it appears to move correctly for the players viewpoint). I should be doing a million other things but was more fun. https://t.co/uGoqwAWWxz

PyTorch 2.11 features improvements for distributed training and hardware-specific operator support. Join Andrey Talman and Nikita Shulga on Tuesday, March 31st at 10 am for a live update and Q&A. Topics: - Differentiable Collectives for Distributed Training - FlexAttention: Now includes a FlashAttention-4 backend on Hopper and Blackwell GPUs - MPS (Apple Silicon): Comprehensive operator expansion - RNN/LSTM GPU Export Support - XPU Graph Register: https://t.co/eJUPu4mCzD #PyTorch #OpenSource #AIInfrastructure

It’s now easier to find, reuse, and build on the files you upload and create in ChatGPT. You can quickly reference files in a chat using recent files in the toolbar, ask ChatGPT about something you’ve uploaded, or browse your files in the new Library tab in the web sidebar. Rolling out globally for Plus, Pro, and Business users, and coming soon to users in the EEA, Switzerland, and the UK.

New Anthropic Fellows research: Automated audits like Petri are increasingly used for alignment evals, but they're often unrealistic, and frontier LLMs can often tell. We measure and improve the realism of agentic coding audits by grounding them in real deployment data. https://t.co/R1BIZOJrtI

Introducing the Anthropic Science Blog. Increasing the pace of scientific progress is a core part of Anthropic’s mission. The Science Blog will feature new research and stories of how scientists are using AI to accelerate their work. Read the intro: https://t.co/1P9BDyX3xG

We’re launching with two new posts. Can AI do theoretical physics? Harvard physicist Matthew Schwartz led Claude Opus 4.5 through a graduate-level calculation. AI can’t yet do original work autonomously, but it can vastly accelerate it. Read more: https://t.co/UUfFuLqhb7

Models keep improving on long-horizon tasks, but splitting work across many agents doesn’t suit every problem. We walk through the setup for a single agent working sequentially on a task where mistakes compound: modeling the early universe. Read more: https://t.co/7rrQY2vp9H

PyTorch 2.11 is now available, featuring 2,723 commits from 432 contributors since PyTorch 2.10. This release prioritizes performance scaling for distributed training and next-generation hardware architectures. Highlights include a FlashAttention-4 backend for FlexAttention on Hopper and Blackwell GPUs, Differentiable Collectives for distributed training, and performance optimizations for Intel GPUs via XPU Graph. This release also delivers comprehensive operator expansion for Apple Silicon (MPS) and RNN/LSTM GPU export support. 🖇️ Read the PyTorch 2.11 release blog and release notes: https://t.co/JZ4xkjEiNQ #PyTorch #OpenSource #AIInfrastructure

The GitHub Security Lab Taskflow Agent is designed to detect Auth Bypasses, IDORs, Token Leaks, and other vulnerabilities that often slip through standard tooling. Want to scan your own project? The taskflows are now open source and easy to run yourself. ⬇️ https://t.co/U5EINWPsL5

https://t.co/yK7QBpOAsb

https://t.co/yK7QBpOAsb

Another privacy implication not mentioned here: recall that on the day Mark Zuckerberg testified during a lawsuit about Meta's alleged harms to underage users, the judge warned that observers could be held in contempt for wearing smart glasses. https://t.co/zEviqIfhl8 https://t.co/I31zVSMMjH

First encounter with a Sturmvogel! Animated with @grok imagine https://t.co/WiCR11UAq3

Grok Imagine generates the best realistic videos that feel truly lifelike ✨ https://t.co/u5oKn0MJq9

1/ Introducing Datapoint AI Intelligence is being commoditized. Good judgement is still rare. We collect human preference data that trains frontier models, by replacing mobile ads with preference tasks. https://t.co/N2AyyzG0VT