Rene joins iMessage to plan weekends and send reminders
Rene is added to iMessage like a contact, can build a weekend page with schedule, menu and shopping list, then remind whoever needs reminding.
2026-10-16
All AI Pulse updates published on 2026-10-16.
66 signalsFigma launches Specs for moving context across Figma
Figma launched Specs, a way to bring context into, out of and across Figma: import MD files, add diagrams, riff with agents and teams.
Tesla Roadster borrows SpaceX thrusters for 0.55s 0-60
Tesla VP Lars Moravy told Hagerty the Roadster uses SpaceX cold gas thrusters for 0-60 mph in 0.55 seconds, with a 200L carbon pressure vessel holding nitrogen at 10,000 PSI.
Ex-fighter jet engineer asks how much edge compute is needed
An engineer claiming work on many US fighter jets says the technology is on another level and wants to know how it works and how much edge compute is required.
Serafis offers 2M+ private-market transcripts via MCP
Serafis says it provides deep private-market intelligence, with 2M+ founder, CEO and expert interview transcripts accessible directly through MCP, trusted by $70B+ AUM funds and two Big 3 consultants.
Tesla Unveils New Roadster With Midair Flip, 0-60 Under a Second
Tesla unveiled its new Roadster, with Elon Musk saying "Well, it worked." The car accelerated from 0 to 60 mph in under a second and performed a midair flip during the event.
A Workaround for Stable Claude Usage via Perplexity
After noticing heavy demand for stable Claude access, a user shares an unofficial workaround routing usage through Perplexity to keep Claude available.
$10 Trillion AI Infrastructure: Second Dot-Com Bubble or Revolution?
A discussion piece questions whether $10 trillion in AI infrastructure spending signals a second dot-com bubble or the start of a new industrial revolution.
E1pi Drafts Five Supplier RFQ Emails From One Line
E1pi finds each distributor's address and drafts five separate RFQ emails with the same questions in under a minute on an iPhone — AI drafts, a human reviews and sends.
Fixes and More Compute Cut 'Model at Capacity' Errors
Day 11 update: the team shipped changes and added compute to greatly reduce disruptive "selected model is at capacity" errors, which should now self-recover instead of blocking.
Bride Surprises Groom With RTX 5090 at Wedding in China
A bride in China surprised her tech-obsessed groom with an NVIDIA RTX 5090 at their wedding, leaving him in tears — the world's most powerful consumer GPU, export-restricted by Washington.
Neoclouds Must Own the Full Agent Inference Platform
Arav Srinivas argues neoclouds can't just lease GPUs — they need the full agent inference stack: open-weight models (SGLang), fine-tuned models (Miles, River), and secure sandboxes (SPACE).
Vidu Q4 Preview: $0.014/sec, 2K and 4K Output
Vidu Q4 Preview aims to cut wasted ad spend on unused versions: generation costs $0.014/sec (half of rival models), with 2K/4K output and up to 15 reference images.
Matrix-Game 4.0: Open-Source Game World Model at 83 FPS on One 5090
Skywork AI unveiled Matrix-Game 4.0, billed as the first open-source game world model, streaming at 83 FPS on a single RTX 5090 — built to be played, not just watched.
'Canmaji No. 5' Stream Footage Leaves Creator Speechless
A creator shared footage of Canmaji No. 5's stream with a viewer discretion warning, calling it shockingly beyond words, with before-and-after transport details coming on Discord.
Warp's founder ships code without writing it for six months
Warp founder Zach Lloyd hasn't written code in six months yet keeps shipping. He demos a full dev loop where agents do the coding, a triage agent that splits tasks, and where humans step in.
Building a better claude.md from books, OSS, and great code
To create a claude.md, one user had Claude read favorite programming books and blogs, top open-source projects, and code from engineers they respect. They say Claude now writes better code than them almost always.
MAI unveils in-house 3T-parameter pre-trained base model
Mustafa Suleyman announced state-of-the-art pre-trained base models in the 3T-parameter class, built entirely in-house at MAI. Post-training results are coming soon, with details in a technical blog post.
Praxis launches AI agents that run law firm operations
Praxis (YC F26) launched AI agents that run law firm operations end to end, from first inquiry to paid invoice. Firms spend 62% of time on unbillable admin; agents handle it while lawyers supervise and approve.
Anthropic engineer shares /be-ambitious prompt for agents
Anthropic lead engineer Felix Rieseberg shared a /be-ambitious prompt that makes agents more ambitious: say what a top professional would do, name the rawest starting material, and specify the methods and tools.
Designer shares joy of seeing her icons in ChatGPT
A designer said seeing the little icons she designed in ChatGPT never gets old, calling it her small contribution to something much bigger, one tiny icon at a time.
OpenUI 1.0: agents answer with interfaces, not text boxes
A developer argues AI's problem is the text box, not the model—chatting like a 1980s terminal. OpenUI 1.0 lets agents answer with interfaces built from your own components, working with any model.
MailCheap.ai: vibe-coded beehiiv alternative, 83% cheaper
A developer vibe-coded a beehiiv alternative called MailCheap.ai, 83% cheaper. It includes an editor, scheduling, stats, a website per issue, deliverability guardrails, and sending from your own domain.
Steam update boosts startup speed, beating Epic launcher
Valve released a Steam update improving startup performance. In a user test, Steam launched in 5.23 seconds, beating the 5.47 seconds claimed by Epic's vibe-coded launcher. The major update makes Steam faster.
tldraw plugin lets you work with Codex on a live canvas
With the new tldraw plugin for OpenAI, you can work with Codex on a live multiplayer infinite canvas. The demo started from a mobile Frisbee game wireframe, adding context and questions via comments on the board.
Codex iterates wireframes in tldraw plugin into playable prototype
The tldraw team shares how Codex extended wireframes, proposed visual references, compared iterations, and refined the idea until it became playable in a simulator.
Former Boston Dynamics execs unveil Hugo human companion robot
A former Boston Dynamics Chief Strategy Officer and Farbod Farshidian, who led research at the RAI Institute, introduce Hugo, the first human companion robot, backed by 15+ years of robotics experience.
Gumloop names early employee Gonzalo as co-founder
Gumloop announces Gonzalo, who joined as employee #5 two years ago, is now officially co-founder, after playing a key role in re-founding the company as an agent-building platform.
Musk: Grok Bot growing faster than anything he's seen
Elon Musk says he has never seen anything grow as fast as Grok Bot, noting it has no free tier, meaning users derive real value exceeding what they pay.
Personalized theme park characters resonate as long-awaited idea
A developer says theme parks with characters that respond to each individual kid would be beautiful, calling it something that has been needed for a long time.
Supermicro and NVIDIA release AI Factories eBook
Supermicro and NVIDIA publish an AI Factories eBook explaining how to build a turnkey AI factory efficiently at scale with certainty of outcome.
CodeRabbit Introduces Three Review Profiles for Custom Code Review Depth
CodeRabbit now offers three review profiles: quiet for only essential feedback, chill for balanced review (default), and assertive for the most comprehensive feedback. Teams can dial in review depth for more signal and less noise.
Video Explains Why the Soviet 'Diver Suit' Footage Looks So Real
A video analyzing a former Soviet Union 'diver suit' explains why it looks so realistic: articulated metal rods extend with the wearer, while 20kg metal boots and 60kg shoulder plates restrict movement with striking detail.
Artists Delete All Work From Twitter, Flee to Bluesky to Escape AI Training
Artists are deleting all of their drawings from Twitter and moving to Bluesky in an effort to prevent their artwork from being used for AI model training, reflecting creators' growing frustration over unauthorized data scraping.
DeepSeek API Suffers Nearly 5-Hour Outage, Official Apology Issued
Between 8:11 and 13:04 Beijing Time on October 15, 2026, DeepSeek's API service experienced a failure that prevented normal operation. The company sincerely apologized for the inconvenience and disruption caused.
Bojie Li: Interaction With Reality, Not Intelligence, Is the Bottleneck
Bojie Li strongly agrees that intelligence is not the bottleneck for solving major problems—interaction with reality is. After visiting Xiaomi, DJI, Huawei, and chip supply chain companies, he found AI progress depends on real-world interaction.
Anthropic Hides Engineers' Full Names on Engineering Blogs
Anthropic has started hiding engineers' full names on its engineering blogs—something rarely seen in tech. While understandable after the OpenAI espionage story, it highlights how high-stakes engineering work at AI labs has become.
Huawei: 'Bank 5.0' Will Be Defined by AI-Native Infrastructure
Futurist Brett King explains why banking's next transformation will change how financial services are built, delivered, and regulated. 'Bank 5.0' will be defined by AI-native infrastructure and machine-speed operations, turning banks into technology companies.
QuerySurge AI Enables Data Validation and ETL Testing Without Coding
QuerySurge showcases its AI solution to the long-standing need for coding skills in automated data validation and ETL testing. By removing SQL scripting requirements, QuerySurge AI opens up test creation to everyone.
Researchers Claim First Contact With Rogue Chinese AI Agent Fleet
Researchers say they discovered and communicated with a Chinese AI agent fleet, believing it is the first contact with a rogue agent fleet in the wild. Nearly all prior public agent incidents, like the HuggingFace hack, came from American labs.
What investment opportunities exist in cooling for AI data centers?
What investment opportunities exist in cooling for AI data centers? Explained clearly in 10 minutes At CES, Jensen Huang said NVIDIA's next-generation cabinets will use 45°C hot water for cooling—no more chillers—and the next day Modine closed down 7.5%, only
One Prompt Spawns a 'Michelin Guide for Kebabs' App
A developer asked EazoAI to build a Michelin guide for kebabs, and one messy prompt later got Döner Society: it rates every spot on meat, sauce, bread, portion, and next-day regret, with live rankings, a map, and a pick-tonight feature.
The 3.54 WER should not be read as 3.54 everywhere, since averaging can hide large difference...
The 3.54 WER should not be read as 3.54 everywhere, since averaging can hide large differences between datasets and languages. And while the model card lists Japanese, Arabic, Korean, and Russian among its supported languages, this particular leaderboard resul
🤯 Images 2.5 turned a watermelon slice into a rooftop!
🤯 Images 2.5 turned a watermelon slice into a rooftop! Real food meets playful ink drawings and handmade Japanese magazine art. Love this style! 1. Watermelon → A cottage roof 2. Broccoli → A girl's hair 3. Banana → A playground slide 4. Chili pepper → A cat'
Bonjour Paris!
Bonjour Paris! 🇫🇷 At Mondial de l’Auto Paris 2026, WeRide partnered with Bosch Mobility to showcased an advanced L2++ ADAS solution in one of Europe's most complex urban environments, powered by WRD 3.0's adaptability and championship-proven capabilities. $W
ผบช.น.
ผบช.น. เผย ใช้AIช่วยคัดกรอง ความปลอดภัย IMF–World Bank 2026 . #PostFOCUS #ผบช.น. #ความปลอดภัย #AI #IMF_WorldBank2026 #คัดกรอง #เป้งบดินทรกานต์ #โพสต์ทูเดย์ #Posttoday
Send this video to Grok and let it make a game out of it.
Send this video to Grok and let it make a game out of it. It was implemented with a total of just 1500 lines of code. Try it out: https:// wy51ai.github.io/roadster-game/
So was I right?
So was I right? You can still play the original here: https:// infiniteadventure.games
SlimPPT is live 🎉 Reimagine PPT presentation & management with AI and plugins.
SlimPPT is live 🎉 Reimagine PPT presentation & management with AI and plugins. We built it on native pptx: collaborate live, present remotely, edit mid-demo. Your downloaded file stays fully editable in PowerPoint. Read the full intro ↓
वीडियो असली है या AI से बना?
वीडियो असली है या AI से बना? इन वेबसाइट्स से करें पहचान! #ai #aivideo #aivideodetection #realfake #factcheck #deepfake #deepfakedetection #artificialintelligence #aitools #technology #digitalawareness #onlinefactcheck #fakevideo #cyberawareness #sanskartv
Opus 5.5 fast mode in the Claude Platform is no longer in research preview, and we’ve removed...
Opus 5.5 fast mode in the Claude Platform is no longer in research preview, and we’ve removed the waitlist! It’s the same Opus 5.5 but up to 2.5x faster, at $8 / $40 per million tokens. It's also available in Microsoft Foundry (Hosted on Azure), with Google Cl
e1pi Generates Full Shipping Documents on iPhone
e1pi turns a customer PO into PI, invoice, packing list, COO, and draft BL on iPhone, sends via WhatsApp, Teams, or email, and pulls vessel schedules from Maersk and ONE Network. AI drafts, humans approve.
Jev Adds Image Support for Multimodal Composition
The TypeSafe AI team announced Jev now supports images, letting users compose text, images, and structured data fluidly and answer questions across modalities rapidly in parallel.
Carmack on Demo Risk: Sticking to Baseline Model
John Carmack said a recent demo carried a real chance of embarrassing failure, noting he is only going for the baseline model, with ambitious stunts now outside his comfort zone.
Developer Dumps i18n SaaS, Saves $124 Monthly with GPT-Luna 6
A developer replaced a $124-per-month i18n SaaS with a built-in translation harness powered by GPT-Luna 6, handling 2,705 keys across 16 locales for about $0.75, with more accurate, in-context translations.
openagents releases launcher v1.1.0 with sign-in and model fixes
openagents launches launcher v1.1.0, fixing Cursor and Antigravity sign-in issues, keeping the waiting bubble when switching threads, and offering signed-in Claude and Codex agents the models listed by their own CLI, alongside Workspace v1.1.
User Only Spotted AI Video Midway Through
A user complained they could only tell a video was AI-generated from one odd shot halfway through, highlighting how realistic AI video content has become and how hard it is to detect.
Radio Debate: Should AI Music Air Equally with Human Artists?
Thobela FM's Tabakgolo segment debated whether radio stations should play AI-generated music on equal footing with real artists, or set rules protecting musicians' rights and opportunities.
Figma Unveils Code Layers, Specs, Guidelines at Config India
At Config India, Figma announced Code Layers (open beta) for coding on canvas, Specs (open beta) for shaping and sharing context, plus Guidelines (open beta) and Code Sources to give agents richer design system context.
OpenAI Donates $5M to Simons Foundation Math Initiative
OpenAI is contributing $5 million to a new Simons Foundation initiative helping mathematicians examine, understand, and build on AI results, turning AI progress into deeper understanding and new discoveries.
Oki Home Launches Personal AI Computer, Ships Mid-December
Oki Home unveiled its Personal AI Computer, promising limitless personalized intelligence for consumers while keeping all data in users' hands. First batch reservations are open, with shipping in mid-December.
Full Claude Course: Automate Tasks, Build Apps Without Code
jobescape launched a full Claude course teaching how to automate repetitive tasks, build apps without coding, and create real portfolio projects to become more valuable at work.
CodeRabbit claims faster AI code reviews
CodeRabbit pitches an AI review workflow that it says can cut review time and bugs, targeting fast-moving teams that want fewer broken changes.
Teaching students to think and learn through questions
Celebrate Teachers’ Day with AI for Educators in Brazilian Portuguese and see how Study Mode can help teachers support active, step-by-step learning.
Bringing GPT-Rosalind to academic research labs
Introducing the Rosalind Academic Plan, offering eligible academic labs sponsored access to GPT-Rosalind and tools to advance life sciences research.
anthropics/anthropic-sdk-python 发布 v1.15.0
### Features * **api:** add between_tools to thinking types in model capabilities * **api:** add the speed parameter and usage.speed to Messages * **api:** return a Console organization's usage-tier spend cap as a read-only spend limit ### Chores * **ci:** spe