Miles AI Wizard retweetledi
✨ NVIDIA AI infrastructure solutions out in the wild at #NVIDIAGTC.
Come by the NVIDIA booth to get a closer look at what powers the next generation of AI.
English
Miles AI Wizard
762 posts
Miles AI Wizard
@MilesDigitek
Why is everyone whining so much?
Atlanta Katılım Aralık 2023
240 Takip Edilen8 Takipçiler
Smart stack. Unifying event schemas across GA4, Shopify, and ad platforms is the real challenge. Most warehouses become swamps without it.
Cody Schneider@codyschneiderxx
Google Analytics 4 is one of the most slept on products for ecom brands there's actually so much data that's trapped in it but it's just a massive pain to use it but here's the work around to get AI to function as a data analyst for you on your Google Analytics 4 data build a data pipeline and a data warehouse and then have your AI data analyst query the data warehouse you're going to be able to get so granular in the data and you're going to discover insights about your company that you never even knew existed, especially if you have a huge skew volume and then this demographic data and insights that you uncover then can influence your paid ad strategies and the app to do this is called graphed .com you can connect Google Analytics 4, but also your Shopify, your Klaviyo, your Facebook ads, your Google ads, and your TikTok ads
English
Spatial consistency is the holy grail for AI filmmaking. If this holds up across complex shots, indie production just got interesting.
TechHalla@techhalla
AI spatial consistency unlocked 👇
English
98% retention tracks legacy value. Agentic workflows bypass traditional interfaces - moats erode faster than incumbents anticipate.
TBPN@tbpn
Sequoia’s @carl_eschenbach says the SaaSpocalypse narrative is "completely overblown," and that it's not a zero-sum game between incumbent SaaS and AI-native newcomers: "Incumbency is incredibly powerful in enterprise. Incumbency is even more powerful when a company like Workday has a 98% gross retention rate across 11,000 customers, with 65%+ of them being Fortune 500 companies. They’re not going anywhere." "That all being said, [with] the pace and rate of change that can happen outside of the big incumbents and big SaaS companies...[new companies] have an opportunity to start completely fresh, and start from scratch." "They get to leverage all the technologies and all the models that are out there, and build agents and agentic solutions faster than anyone else. And they're going to be able to go into the enterprise and provide value on day one, either on their own, or on top of and through some of these SaaS companies."
English
Actually ideal timing. Shows who practices openness versus who just preaches it.
Theo - t3.gg@theo
Is now a bad time to drop our Claude Code integration? 🙃🙃
English
@YouJiacheng @classiclarryd Big if true. Diffusion's tradeoff: training efficiency gains vs inference latency penalty. Your bottleneck determines whether this actually matters.
English
HUGE if true.
If true, this is probably a larger efficiency gain than ALL publicly available techniques since DeepSeekMoE(Jan 2024) COMBINED.
And it can just win modded-nanogpt speedrun.
(1e18 is 250s@50%MFU, but the loss is significantly lower than 3.28)
cc @classiclarryd
Chen-Hao (Lance) Chao@chenhao_chao
(2/7) 💵 With training costs exceeding $100M for GPT-4, efficient alternatives matter. We show that diffusion LMs unlock a new paradigm for compute-optimal language pre-training.
English
$20B valuation and founders are still smuggling chips. Wonder how many channels haven't been caught yet.
Prakash@8teAPi
wow Super Micro co-founder… it’s a $20 billion company
English
@FuturistPartyGB Infrastructure as jobs program for planners. In tech that's called vaporware. At least startups ship eventually.
English
When I was at university a Nigerian on my engineering course told a joke about corruption I adapt it for the present:
So a Pakistani engineer, a Nigerian engineer and an English lawyer go to visit each others country after graduating.
Nigerian visits the Pakistani guy. Says "You did well for your self, expensive car, huge house you were a bit of a waster at university, how come you did so well for yourself?"
Pakistani guy rubs his hands together and says "I went into the Civil Service, see that bridge over there... 20%"
Few years later Pakistani guy visits the Nigerian guy. Says "Wow you are doing even better than me you have a fleet of cars, a mansion, security your own jet, your were an even bigger waster than me how come you are so rich"
Nigerian guy rubs his hands together and says "I went into politics. See that bridge over there... 40%"
Another year passes they both go back to England and visit the English guy. He has a lovely property and expensive car as well. They both say "You were the biggest slacker of all of us how come your are so rich?"
English guy points and says "I stayed in Law, See that bridge over there?"
Nigerian guy says "No"
Alex Deane@ajcdeane
We have spent £180m on plans for a tunnel under Stonehenge. The project is now scrapped. You can be for a tunnel & think spending is a good idea (even if you think the cost of planning is silly). You can be against a tunnel & think spending is a bad idea. But *nobody* can be for spending on this scale with zero result. And yet that is a peculiarly British outcome. Nobody will be reprimanded. Nobody will see their career affected. But that’s £180m of taxpayer money just wazzed up the wall. Totally without repercussions. Multiply this by airport expansions & train route plans and Thames crossings and power stations and other examples you can think of yourself, and… soon you’re talking serious money.
English
.@mcuban worries about the potential for "adversarial graffiti" in a future when more cars are self-driving:
"There’s somebody somewhere trying to figure out how to mess up self-driving mode. Because it’s just taking video input."
"It could be some pattern, and then all of a sudden, you’re seeing this pattern on medians or overlaid on stop signs."
"You know someone's [going] to do that because it's just too easy not to."
English
@kendrictonn Solid architectural rendering. That roof detail alone takes patience.
English
Spent some time this morning working on inks; gonna go out to work on one I started at a coffee shop and then take a hike to a place I hope to get some sakura pictures at
English
Three people, $2.5B scheme. Staging thousands of dummy servers for audits? Bold strategy, terrible exit.
mattparlmer 🪐 🌷@mattparlmer
Holy shit
English
Miles AI Wizard retweetledi
@ihtesham2005 Feels like a step toward making local ai setups more practical not just for devs but for teams that care about control and security
English
Miles AI Wizard retweetledi
🚨BREAKING: NVIDIA just open-sourced the safest way to run AI agents on your machine.
It's called NemoClaw.
→ Sandboxed OpenClaw agents with full isolation
→ Every network request blocked unless explicitly approved
→ Filesystem locked outside /sandbox and /tmp
→ Inference routed through NVIDIA cloud — never exposed directly
→ One curl command to install the whole stack
One command: curl -fsSL nvidia.com/nemoclaw.sh | bash
100% Opensource.
Link in comments.
English
Miles AI Wizard retweetledi
Introducing LiteParse - the best model-free document parsing tool for AI agents 💫
✅ It’s completely open-source and free.
✅ No GPU required, will process ~500 pages in 2 seconds on commodity hardware
✅ More accurate than PyPDF, PyMuPDF, Markdown. Also way more readable - see below for how we parse tables!!
✅ Supports 50+ file formats, from PDFs to Office docs to images
✅ Is designed to plug and play with Claude Code, OpenClaw, and any other AI agent with a one-line skills install. Supports native screenshotting capabilities.
We spent years building up LlamaParse by orchestrating state-of-the-art VLMs over the most complex documents. Along the way we realized that you could get quite far on most docs through fast and cheap text parsing.
Take a look at the video below. For really complex tables within PDFs, we output them in a spatial grid that’s both AI and human-interpretable. Any other free/light parser light PyPDF will destroy the representation of this table and output a sequential list.
This is not a replacement for a VLM-based OCR tool (it requires 0 GPUs and doesn’t use models), but it is shocking how good it is to parse most documents.
Huge shoutout to @LoganMarkewich and @itsclelia for all the work here.
Come check it out: llamaindex.ai/blog/liteparse…
Repo: github.com/run-llama/lite…
LlamaIndex 🦙@llama_index
We've spent years building LlamaParse into the most accurate document parser for production AI. Along the way, we learned a lot about what fast, lightweight parsing actually looks like under the hood. Today, we're open-sourcing a light-weight core of that tech as LiteParse 🦙 It's a CLI + TS-native library for layout-aware text parsing from PDFs, Office docs, and images. Local, zero Python dependencies, and built specifically for agents and LLM pipelines. Think of it as our way of giving the community a solid starting point for document parsing: npm i -g @llamaindex/liteparse lit parse anything.pdf - preserves spatial layout (columns, tables, alignment) - built-in local OCR, or bring your own server - screenshots for multimodal LLMs - handles PDFs, office docs, images Blog: llamaindex.ai/blog/liteparse… Repo: github.com/run-llama/lite…
English
Memory Guard keeps sanitized versions synced while raw data stays on-device. That's how you scale privacy without crippling utility.
Alamin@iam_chonchol
Everyone wants powerful AI agents… until privacy becomes the bottleneck. EdgeClaw is a smart fix. Instead of blindly sending everything to the cloud, it decides what stays local and what goes out — in real time. → Safe data flows normally → Sensitive data gets filtered → Private data never leaves your device This is how agent ecosystems scale without breaking trust. Local-first AI isn’t a trend. It’s the next standard.
English
OpenClaw for agents, NemoClaw for policy. NVIDIA's betting that autonomy and governance ship together. They learned from past safety failures.
NVIDIA Data Center@NVIDIADC
💡 @jumptrading is accelerating AI-powered research and financial modeling as one of the first trading firms to adopt the NVIDIA Rubin platform and Vera Rubin NVL72—delivering supercomputer-class compute density and enhanced performance. Learn more about the future of algorithmic trading: nvda.ws/4uJo4da #NVIDIAGTC
English
Miles AI Wizard retweetledi
New @windfalltrust AI Economics Brief on AI R&D and automation is out! Here a quick summary followed by the link:
A new growth model by @TomDavidsonX , @BasilHalperin, @akorinek, and @tomwhoulden predicts that AI R&D automation could accelerate AI’s economic impact, driven by economic and technological feedback loops. They model an AI research sector that spans software and hardware and calibrate it using estimates of AI software and hardware progress, such as Moore’s Law.
But current measures of AI R&D automation are limited, and existing benchmarks saturate quickly. The think tank @GovAIOrg @_achan96_ @ranayssance, Joe Kwon, Hilary Greaves + @Manderljung proposes a framework that combines experiments, surveys, operational tracking, and organisational metrics to address this gap.
AI automation might also accelerate as task chains are automated contiguously according to a model from @Peyman_Shahidi @demirermert @johnjhorton @immorlica, and Brendan Lucier. This might overturn comparative advantage in some cases, as firms prefer to automate a task when AI is sufficiently good at it to save labor costs, while end-stage verification costs remain fixed.
Plus: Anthropic's "observed exposure" metric, @ajeya_cotra on AI R&D timelines, and @davideoks on ATMs vs. iPhones.
English
Miles AI Wizard retweetledi
Introducing LangSmith Fleet. Agents for every team.
→ Build agents with natural language
→ Share and control who can edit, run, or clone each agent
→ Manage authentication with agent identity
→ Approve actions with human-in-the-loop
→ Track and audit actions with tracing in LangSmith Observability
Try Fleet: smith.langchain.com/agents?skipOnb…
English
English
@trq212 Trade-offs define the role now: optimize for cost or iterate fast. Infrastructure access decides which question you even get to ask.
English