Noa Shavit (@noashavit) header
Noa Shavit (@noashavit) profile picture

Noa Shavit

@noashavit

💬 Tweet about data, dev tools, startups, and OSS 👩‍💻 data infra products by day, vibe coding agent I use daily after hours

 San Francisco, CA
Joined October 2011
  • Tweets84.6K
  • Following2K
  • Followers2.6K
1.5KMedia
Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)Photo by Noa Shavit (@noashavit)
Why Anthropic's Filing Crushed Tech Stocks thebignewsletter.com/p/why-a…
Anthropic Isn't Just a Company, It's a Payoff

Anthropic Isn't Just a Company, It's a Payoff

The trillion dollar Anthropic IPO will enrich Google and Amazon, who are ostensibly competitors. How does that make sense? It doesn’t, unless Anthropic is four flailing monopolies in a trench coat.

thebignewsletter.com
14
Accenture Just Capped Employees’ Claude Spending at $29 a Month. What exactly can you build with that on API rates? 247wallst.com/investing/2026…
Accenture Just Capped Employees’ Claude Spending at $29 a Month. Some Were Spending $1,000

Accenture Just Capped Employees’ Claude Spending at $29 a Month. Some Were Spending $1,000

Accenture sells AI transformation to thousands of clients while quietly telling its own employees they get a fraction of what some were already spending. The internal memo reveals a tension at the...

247wallst.com
36
Apate sends AI agents undercover to get scammed so they can gather fraud intelligence techcrunch.com/2026/10/14/ap…
Apate sends AI agents undercover to get scammed so they can gather fraud intelligence | TechCrunch

Apate sends AI agents undercover to get scammed so they can gather fraud intelligence | TechCrunch

Most companies use AI agents to ship products faster or speed up workflows. Apate.ai sends them undercover to get scammed.

techcrunch.com
53
TypeSafe AI raises $870M for Jev: decision model that never generates text msn.com/en-us/news/other/typ…
83
Effective Pricing Strategies for AI Coding Agent Subscriptions rswebsols.com/news/strategie…
Effective Pricing Strategies for AI Coding Agent Subscriptions

Effective Pricing Strategies for AI Coding Agent Subscriptions

Discover proven strategies to price your AI coding agent subscription effectively, maximizing profitability while staying competitive in a fast-growing market.

rswebsols.com
139
A good read linkedin.com/pulse/how-ai-ag…
How AI Agents Actually Get Better in Production

How AI Agents Actually Get Better in Production

The recursive learning loop that separates systems that compound from systems that quietly rot, and the one thing that decides which you get. I build AI systems for a living, and for the last year I...

linkedin.com
36
Meta employees consumed 73.7 trillion tokens in 30 days- crazy! mlq.ai/news/meta-caps-intern…
Meta Caps Internal AI Token Spending After Costs Approach Billions in 2026

Meta Caps Internal AI Token Spending After Costs Approach Billions in 2026

Meta is imposing spending controls on employee AI usage after internal costs approached billions of dollars in 2026, according to an internal memo reported by The Information. CTO Andrew Bosworth tol…

mlq.ai
32
Reliable data infrastructure means assuming failure. Backup your Postgres with this OSS project: github.com/EnterpriseDB/barm…
GitHub - EnterpriseDB/barman: Barman - Backup and Recovery Manager for PostgreSQL

GitHub - EnterpriseDB/barman: Barman - Backup and Recovery Manager for PostgreSQL

Barman - Backup and Recovery Manager for PostgreSQL - EnterpriseDB/barman

github.com
36
See inside the black box of GPU utilization! github.com/systalyze/utilyze
GitHub - systalyze/utilyze

GitHub - systalyze/utilyze

Contribute to systalyze/utilyze development by creating an account on GitHub.

github.com
20
You can always have Claude code use your local Ollama modesl, but here's an easy to add them to the UI too github.com/aattaran/deepclau…
GitHub - aattaran/deepclaude: Use Claude Code's autonomous agent loop with DeepSeek V4 Pro,...

GitHub - aattaran/deepclaude: Use Claude Code's autonomous agent loop with DeepSeek V4 Pro,...

Use Claude Code's autonomous agent loop with DeepSeek V4 Pro, OpenRouter, or any Anthropic-compatible backend. Same UX, 17x cheaper. - aattaran/deepclaude

github.com
22
Mistral is the European Claude: Le Chat is Claude Chat and Vibe is Claude Code. thenewstack.io/mistral-vibe-…

Mistral, Europe’s answer to OpenAI and Anthropic, pushes its coding agents to the cloud

Paris-based AI startup pushes its Vibe coding agent into the cloud, launches a new model, and keeps the heat on the major AI labs.

thenewstack.io
45
The exec that never opened Claude...
Jaya Gupta (@JayaGup10) profile pictureJaya Gupta
@JayaGup10Apr 24

Experience is now a tax.

Article · @JayaGup10
1
195
The biggest AI infrastructure trend is the move to classified, air-gapped deployment. Security and governance are the new compute constraints. techcrunch.com/2026/05/01/pe…
Pentagon inks deals with Nvidia, Microsoft, and AWS to deploy AI on classified networks | TechCrunch

Pentagon inks deals with Nvidia, Microsoft, and AWS to deploy AI on classified networks | TechCrunch

The deals come as the DOD has doubled down on diversifying its exposure to AI vendors in the wake of its controversial dispute with Anthropic over usage terms of its AI models.

techcrunch.com
23
Local alternative to Claude Design. Coding agents need to do more than just write code. Their output should include actionable design. This OSS project caught my eye- worth a try! github.com/nexu-io/open-desi…
GitHub - nexu-io/open-design: The 1st Collaborative AI Design Agent Workspace. Local-first and open...

GitHub - nexu-io/open-design: The 1st Collaborative AI Design Agent Workspace. Local-first and open...

The 1st Collaborative AI Design Agent Workspace. Local-first and open source. Create motion, prototypes, websites, apps, dashboards, presentations, images, and videos with your team and AI agents. ...

github.com
93
The fight over AI tooling isn't about the model anymore. It's all about the distribution layer. The platform that controls the developer experience wins. It's a battle for the IDE.
9
Local AI data analysis might be the future of data exploration. Local analysis means nothing ever leaves your machine. Saving the process as a notebook preserves the context, the assumptions, and the steps- all critical for auditability. mljar.com/
MLJAR Studio – AI for Data Analysis

MLJAR Studio – AI for Data Analysis

MLJAR Studio is a private AI data lab for exploring data, running machine learning experiments, and building analysis tools. Runs locally with optional AI providers.

mljar.com
11
The AI compute race extends to real-estate techcrunch.com/2026/05/01/co…
Coatue has a plan to buy up land for data centers, possibly for Anthropic | TechCrunch

Coatue has a plan to buy up land for data centers, possibly for Anthropic | TechCrunch

Coatue, one of the biggest names in venture capital, has a new venture that is reportedly buying land near large power sources.

techcrunch.com
13
AI should be portable. I use local models all the time with Ollama. Came across this today getclaw.site/
claw — Tiny POSIX-sh Shell Agent for any Linux box

claw — Tiny POSIX-sh Shell Agent for any Linux box

A single shell script. Streaming chat against OpenAI or Anthropic. Shell tool calls, rolling memory, skills, mentor mode. No Node, no npm — just curl + jq.

getclaw.site
11
Building agents shouldn't require a PhD in prompt engineering. Frameworks like LangGraph let us treat agentic workflows as structured, testable code. Focus on the graph, not the grammar. langchain.com/langgraph
LangGraph: Agent Orchestration Framework for Reliable AI Agents

LangGraph: Agent Orchestration Framework for Reliable AI Agents

Control agent workflows with LangGraph: durable execution, memory, streaming, and human-in-the-loop.

langchain.com
33
State of the art of coding models, according to Hacker News commenters. Refreshes every 24 hours. hnup.date/hn-sota

HN SOTA — Model popularity | HN Update

LLM model popularity from Hacker News: how we detect models from the OpenRouter catalog and score sentiment from discussions.

hnup.date
17
Load more