Andy Xu

Vijay Bendigeri, Ph.D., excellent post. The truth is, this new multi-trillion-dollar Industry requires INTELLIGENT INFRASTRUCTURE to address supply chain logistics, optimization, orchestration, reduced emissions, Vision Zero, and enabling Autonomy 2.0. Exceptional work has advanced the vehicle with sensors, software, computing, AI algorithms, and safety features. It is far more than the vehicle. Exciting times for our nation, Building Our Next Golden Age. The national buildout of Intelligent Infrastructure Economic Zones will revitalize infrastructure in the United States. Supporting NextG for DoD and Civil, FAA NextGen, Intelligent Transportation, Golden Dome, resilient grids, precise positioning, and autonomous systems. ODC AI-RAN is the convergence of AI and the physical world. Autonomy Institute https://lnkd.in/giehikzr

7

This is interesting: the smallest #Sail instance on ClickBench using c6a.2xlarge (8 vCPUs and 16 GiB of memory) outperforms every #Spark and Spark accelerator configuration across all instance sizes. The best-performing Spark setup is the Comet Accelerator on c7a.metal-48xl (192 vCPUs and 384 GiB of memory). It’s wild that Sail, running on just 8 vCPUs and 16 GiB, still significantly outperforms an accelerated Spark workload running on 192 vCPUs and 384 GiB. Notably, both Sail and Comet are powered by #DataFusion. Benchmark: https://lnkd.in/gSMjxrtR Sail: https://lnkd.in/gQiakuU6

66

1 Comment

TL;DR: The best way to beat a viral open-source project? Build the same thing "better". And ship it inside the product people already pay for. Welcome to Claude Code Channels. That's exactly what Anthropic just did. This week, Anthropic launched Claude Code Channels: a native way to connect Claude Code directly to Telegram and Discord. You assign it a task, step away, and get a message when it's done. Your phone becomes a remote control for your entire dev environment. This is a direct answer to OpenClaw, the open-source agent that became the fastest-growing project in GitHub history by offering this exact workflow. OpenClaw was compelling. It was also a security minefield: a CVSS 8.8 vulnerability, 50,000+ exposed instances, and a community skills registry where ~12% of plugins turned out to be malicious. Anthropic watched all of this unfold and responded not with a legal strategy (well, also that), but with a product strategy. The deeper shift here isn't about Discord bots or Telegram integrations. It's about a fundamental change in how we work with AI:  • From synchronous (ask → wait → answer) to asynchronous (assign → go live your life → receive result)  • From AI as a tool, you visit to AI as a collaborator that runs in the background  • From open-source workarounds to enterprise-grade defaults Developers have been hacking together "always-on AI" workflows for months because the demand was real. Anthropic just made it native, sandboxed, and safe. The question for every team building on AI right now: are you still designing for synchronous interaction, or are you building for the async-agent world that's already here? #AI #DeveloperTools #ClaudeCode #ArtificialIntelligence #SoftwareDevelopment

4

3 Comments

Open AI’s Codex 5.3 and Claude Opus 4.6 (released on Feb 5th) aren’t just incremental upgrades, they signal a structural shift. What’s emerging is not better content generation. It’s increasingly dependable multi-step reasoning inside real workflows. And that is a big jump. When models can plan, execute, and maintain context with growing stability, AI stops being a tool you use. It becomes a system you design around! #AI #GenerativeAI #LLMs

38

3 Comments

MiniCPM5-2B is a dense 2B open-source SOTA model optimized for on-device deployment, local coding agents, and agentic workflows, featuring a 131k context length and accompanying high-quality training datasets. https://lnkd.in/gUes558T #MachineLearning #AI #LLM #DeepLearning #AgenticAI

1

One nice tip for your AI workflows....NotebookLM has a great capability for turning a pile of messy unstructured information from documents or websites into a neat "data table", ie structured, organized data that is easily ingestible in AI orchestration with high quality and low-latency/overhead. This has come in handy for me more than once so I thought I'd share in case it helps someone else who's trying to organize input data quickly for their AI features.

15

1 Comment

So if semantic objects aren't schema objects like tables and columns, what are they? In short, they are units that link a concept in the real world to data in the database when those concepts need to be modeled as logic rather than as a column. That's right -- semantic layers are only required because database schemas are rigid, slow-moving, unversioned and unversionable objects. That's why people who only work on toy datasets like taxis and iris can't understand why a semantic layer would be necessary. IT ISN'T...if your data model is perfect. But in the real world, the best planned data model gets out of data almost immediately. Even if one source system manages to reflect reality perfectly, introducing a new data source by definition introduces a conflicting data model. If you want to query across them you either have to create a third harmonized data model and perfectly maintain it...or create a semantic layer that abstracts the objects they share into a parallel plane. And in that plane, we can be flexible, move fast, use versioning...all without weeks of ELT writing. So, finally, what is in this plane? We haven't normed on this as an industry, but I have an opinionated view, which is the view we've built http://schemantic.io around. The core objects are: - Entities - Events - Attributes ## Entities This is the foundation of the semantic layer. Each one of these represents a distinct noun in reality -- people, places, things, ideas. Fundamentally, these are ID columns. Everything else "hangs" on these objects. ## Events If entities roughly represent space, events represent time. They define how and when entities interact. They are fundamentally dates, and their names explain what the interaction is. Because of this principle, events "link" entities together. ## Attributes Virtually everything else in a database is an attribute. These contain the characteristics of an entity. Fundamentally, these values can be integers, strings, booleans. They can be categories, scalars, anything at all. They just need to be "bound" to an entity. Each of these objects has its own logic, written in SQL, so that they can composed by humans and machines into the infinitely complex queries we all need to write. And like all objects (both schema and semantic), they have context. But that is a topic for my next thread. If you want to see how a semantic layer can accelerate your analytics and AI readiness, request a demo here: https://lnkd.in/gXyjiq3e

14

2 Comments