All AI Pulse updates published on 2026-06-28.
35 signalsCOMPUTE AT THE EDGE IS NO LONGER A LUXURY Jensen Huang just dropped the Jetson Nano, and the industry standard for AI inference efficiency has shifted. The unit economics: $249 entry cost. 25W power envelope. 70 trillion operations per second. This hardware is
FT: Google capped Meta’s use of Gemini after Meta asked for more model compute capacity than Google could supply. Meta’s problem is that it uses Gemini inside safety automation, customer support, ad tools, coding, and internal workflows. Google’s problem is di
Google Brain founder Andrew Ng just said prompting is almost over "100% of my tasks are done by AI agents" "Self-improving loops are next" "Give it 3-6 months and prompting is gone" In 6 minutes, he breaks down how to build agents from scratch: Loops Agents Fe
This Chinese guy built a $12,000 MRR business with his phone and Claude > 0 competition > 6 figs/a month potential > almost $0 expenses Here what he does: 1) He places his phone on a motorized rotating mount 2) Sets it in the middle of a room 3) Walks away The
99% of second brains are fake. They store notes and recall nothing. A real one looks like this. Dense, wired, every note pulling on the ones around it. Ask it anything and the answer is already connected. He didn't build that by hand. Claude files every source
Elon on why a smarter Grok might not be a safer one "I think you need to make sure that Grok says things that are correct, not politically correct" pause at 1:20 - that's the line that reframes this entire debate "Physics is law. Everything else is a recommend
Someone said AI is the biggest financial opportunity for anyone under 25. He didn't scroll past. Look at 0:23. That's a working AI agent for McDonald's. Built in under 60 seconds. He watched it once and repeated it the same evening. He now sells agents like th
he built a talking AI that’s convinced it’s an ancient pharaoh no internet, no subscription, no account - it runs completely offline kids at his sister’s school won’t stop talking about it now he’s quietly building them to order at $200 each on screen it intro
NVIDIA CEO Jensen Huang says claims that AI is taking jobs "COMPLETE NONSENSE."
A DEV TURNED A USED 2019 MAC PRO INTO A 35B LOCAL LLM SERVER BY PLUGGING IN ONE AMD EGPU AND PUSHING THE RIG TO 52 TOKENS PER SECOND FOR UNDER $3,000 TOTAL he posts a video of the mac pro under the desk, one thunderbolt cable running to a Paladin eGPU enclosur
You can build an AI second brain this weekend. No $1000 course. No coding bootcamp. Just your own notes and one free, open-source tool. Here's the whole setup: Step 1 - Put your notes in one folder. Obsidian, markdown, PDFs, old docs. everything, one place. St
CHINESE DEV BILLS $22K A MONTH RUNNING AI ON $80 CARDS BUILT FROM PS5 SILICON WHILE EVERY AI STARTUP SITS ON AN 8-MONTH H100 WAITLIST Lenovo office PC. AMD BC-250 sticking out the side. Same chip Sony ships in a PS5 now running inference for 3 paying clients.
EXCLUSIVE: Anthropic CEO Dario Amodei says he started Anthropic because Altman is liar not because of safety reasons.
This guy built a $5,000 agency-style portfolio site in 2 hours with Claude Code It doesn't look like a template It looks like a dark cinematic website with custom sections, animated transitions and a cursor flashlight effect on the hero image But the workflow
CLUSTERING 24 BC250 BOARDS FROM DEAD CRYPTO RACKS INTO A 384GB INFERENCE POOL HIT 1 TOKEN PER SECOND AND $5,000 ANNUAL POWER, THE TIER ZERO HACK ON YOUR MAP DOES NOT SCALE PAST 4 NODES 01:18 the operator looks at his bench, "it could run the model but you're t
Eric Schmidt, former Google CEO: "If you're not using this technology, you're not going to be relevant compared to your peer groups and your competitors and the people who want to be successful. Adopt it and adopt it fast." In a 25-minute TED talk, he draws th
CLAUDE + 5 LOOP COMPONENTS = THE FASTEST WAY TO BUILD WHILE YOU SLEEP most people use coding agents like chatbots, then wait at every step instead of building a system that finds and completes its own work: > Automation wakes the loop on a schedule or event >
Reuters: Austria is asking the EU to give Anthropic a European base because US AI controls have turned frontier-model access into a geopolitical dependency. It is asking Brussels to explore a political and legal package that could make Anthropic run part of it
Claude Code just turned a 3-hour dashboard build into 1 prompt and a link. Anthropic shipped it June 18. Beta. Team and Enterprise only. Here’s what it does. You work in the terminal. Claude turns that session into a live webpage. A real URL you send straight
THIS DEVELOPER RUNS A FULL LOCAL AI WITH WIKIPEDIA-SCALE KNOWLEDGE BASE - AND PAYS $3/MONTH WHILE OTHERS PAY $300 open-frame AI workstation, LCD display showing real-time GPU load and memory usage, PCIe cards for local inference - and a monitor showing a knowl
A $10,000 GADGET JUST REPLACED THIS DEVELOPER'S KEYBOARD. And somehow... It made Claude Code even faster. No typing. No mouse. Just voice commands and physical controls. Every button launches a different AI workflow. Every knob controls an agent. The display s
THIS GUY SHOWED THE CODEX SUBAGENT FLOW FOR FINISHING APPS 20X FASTER The useful part is not “AI writes code.” It is the shape of the prompt. Most people use Codex like this: one chat, one task, one thread, then you personally remember the UI bugs, product log
BY CONNECTING OBSIDIAN, NOTEBOOKLM, AND HERMES AGENT ON A MAC MINI, HE FIXED AI MEMORY LOSS Using public models without local context is like giving your agent dementia in every new session He solved this by wiring Obsidian as a local memory layer on his machi
A developer just open-sourced an AI company. Not an AI tool. An entire company. 147 AI agents. 12 specialized divisions. 50,000+ GitHub stars in under 2 weeks. Engineering. Marketing. Design. QA. Product. Each agent has its own role, workflow, and personality.
This girl says she makes up to $5k creating cartoons using AI She revealed her “secret” to earning on YouTube with AI, and it honestly looks kind of scary In the video, she breaks down step by step her workflow that turns AI into a personal production studio:
Your AI agent writes 50 lines where a senior dev writes 1. Ponytail drops the laziest senior dev straight into your coding agent. Zero to 60k stars in two weeks. Ask for a date picker. Most agents install a library, add a wrapper, a stylesheet, then start a ti
AIKEK is prepared for the local model revolution with hundreds of specialized models.
A Chinese developer spends $74K monthly on AI using a $900 RTX 3090.
Victoria's Secret pays a model $100,000 for a campaign I built a non-existent model for.
Boris Cherny integrated everything he knows into one loop and let it run automatically.
Karpathy's second brain pattern visualizes notes into a graph connecting ideas to projects.
These companies quietly released core tools on GitHub. Here are 5 repositories.
This developer transformed a $599 Mac Mini into a private AI computer fitting in a backpack.
Box's CEO discusses AI access as a new tool in geopolitics.
This developer showcases every small computer from Raspberry Pi to Mac Mini running local AI.