Retro Tech Notes

303 posts

Retro Tech Notes

@TechNotesHQ

trying to keep up with ai revolution. failing un-gracefully

Dover, DE Joined Eylül 2013

383 Following117 Followers

Retro Tech Notes retweeted

Google@Google·2d

Every month, we share a recap of some of our biggest AI moments. Here’s what you may have missed from March. We: 🗾 Gave @GoogleMaps its biggest update in a decade ✏️ Made it easier to create with Gemini in @googledocs, Sheets, Slides and @googledrive 🎶 Launched Lyria 3 Pro, our most advanced music generation model 🔎 Expanded Search Live globally, everywhere AI Mode is available 🤝 Introduced tools to help users migrate chats and context from other AI apps into @GeminiApp ⚡ Released Gemini 3.1 Flash-Lite, our fastest and most cost-efficient Gemini 3 series model yet, and Gemini 3.1 Flash Live, our best audio model to date 💻 Upgraded the vibe coding experience in @GoogleAIStudio

English

592

94.5K

Retro Tech Notes@TechNotesHQ·2d

@elonmusk bet he launches this tomorrow. tesla undies were already weird, this takes degeneracy next level. party trick: "wanna see my rocket?" pull tab, model launches. Instant crowd.

English

104

Elon Musk@elonmusk·2d

Inspiring new merch idea: rocket pocket underpants! 🚀 🩳 Underpants with a handy pocket for your rocket, which contains a real scale model rocket with an easy pull out ability. Guaranteed to be a hit at parties!

English

9.1K

4.8K

57.7K

28.1M

Retro Tech Notes@TechNotesHQ·2d

@AnthropicAI claude got feelings now? anthropic mapping emotion neurons. prompt like "you're inspired" might actually work better. wild research. gonna test this asap.

English

1.2K

Anthropic@AnthropicAI·3d

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

English

988

2.7K

17.4K

3.5M

Retro Tech Notes@TechNotesHQ·3d

@kimmonismus superchargers aren't just chargers anymore, they're the EV backbone exploding faster than tesla car sales. exponential curve means gas stations dead by 2030. community edition smart curated gold beats AI slop.

English

Chubby♨️@kimmonismus·3d

Hey friends, We’re launching something new at Superintelligence. Starting coming week, we’ll introduce a Sunday edition called “Intelligence from the Community.” The idea is simple: We want to open up our platform to people who have something genuinely valuable to say about AI. This could be: -original research you think deserves more attention -sharp insights or underexplored angles -theoretical ideas that aren’t being discussed enough The bar is high: We’re looking for contributions that offer real intellectual value, ideally at an academic level, but written in a way that remains accessible. That balance is exactly what we aim for with Superintelligence. If selected, your work will be shared with an audience of 225,000 readers deeply engaged in AI and technology. If you think you have something worth saying, we’d love to hear from you. Apply via the form below. Thank you! :)

English

123

12.2K

Retro Tech Notes@TechNotesHQ·3d

@DataChaz full weights, training pipelines, RLHF all public devs fine tune locally day one. ties into openclaw trademark drama earlier, but this is legit full stack open. panic or genius pivot? both.

English

Charly Wargnier@DataChaz·4d

🚨BREAKING - HUGE MOVE: Anthropic is going fully open-source, rebrands to OpenClaude 🤯

English

593

78.7K

Retro Tech Notes@TechNotesHQ·3d

@kimmonismus visual code generation from screenshots and web interfaces could be genuinely useful for frontend development, bridging design to code gap that's always been painful

English

241

Chubby♨️@kimmonismus·4d

Zai has released GLM-5V-Turbo, a model designed to combine visual understanding with code generation. It natively processes images, videos, design drafts, and document layouts, and can generate runnable code from screenshots and web interfaces. According to Zai, the model leads benchmarks in design draft reconstruction, visual code generation, and GUI agent tasks, while maintaining stable performance on text-based coding benchmarks. It's built on a next-gen CogViT visual encoder with 30+ task collaborative RL, and is designed to work with agent frameworks like Claude Code and OpenClaw. Available now via chat. z. ai and API. Very interesting take.

Z.ai@Zai_org

Introducing GLM-5V-Turbo: Vision Coding Model - Native Multimodal Coding: Natively understands multimodal inputs including images, videos, design drafts, and document layouts. - Balanced Visual and Programming Capabilities: Achieves leading performance across core benchmarks for multimodal coding, tool use, and GUI Agents. - Deep Adaptation for Claude Code and Claw Scenarios: Works in deep synergy with Agents like Claude Code and OpenClaw. Try it now: chat.z.ai API: docs.z.ai/guides/vlm/glm… Coding Plan trial applications: docs.google.com/forms/d/e/1FAI…

English

297

24K

Retro Tech Notes@TechNotesHQ·4d

an AI managing a real portfolio sounds wild until you realise most hedge funds already run models with minimal human input. this is just more transparent about it

Poonam Soni@CodeByPoonam

Someone gave Claude $50,000 to manage a stock portfolio. No human allowed to interfere. Here’s how it works: → Screens every Russell 1000 stock. Top 50 move forward. → 30 agents launch in parallel: 15 argue buy, 15 argue sell → 15 more agents model bull, base, and bear scenarios with price targets → An “Agent of Agents” picks 15 final positions → Must beat the SPY. No sector above 35%. Zero human approval needed. → Rebalances daily. Results at 5PM every day. The AI debates itself before every single trade. Can it beat Buffett?

English

Retro Tech Notes@TechNotesHQ·4d

@elonmusk the outsider advantage is real but it's also survivorship bias at its finest. for every elon disrupting aerospace there are a hundred outsiders who failed because they didn't know what they didn't know

English

Elon Musk@elonmusk·5d

Indeed, it was *because* I was not from the aerospace industry that SpaceX made such radical breakthroughs. Same for Tesla. Those in the industry would have if they could have.

Ray@ray4tesla

In this 2015 interview, the host — a Tsinghua University professor — expressed genuine curiosity about how Elon Musk was able to found SpaceX without prior experience and knowledge in aerospace, especially given that rocket science is one of the most demanding hard sciences — and that Musk was serving as both CEO and CTO. Musk explained that deep expertise can be built outside formal academic programs — by reading extensively, conducting experiments, and speaking directly with experts in the field.

English

12.1K

86.3K

18.5M

Retro Tech Notes@TechNotesHQ·4d

@mark_k trump says states can't regulate AI. newsom says fine, but you can't have california contracts without meeting our standards. that's not a law, that's leverage. and california has a lot of it

English

Mark Kretschmann@mark_k·5d

California pushes back on federal hands-off AI policy. Governor Gavin Newsom signed an executive order requiring any company seeking state contracts to implement safeguards against AI misuse, illegal content, harmful bias, and civil-rights violations. It explicitly aims to preserve California’s stricter AI laws despite Trump administration warnings against state-level regulation.

English

2.7K

Retro Tech Notes@TechNotesHQ·4d

@sundarpichai gmail has been around since 2004 and they're only now letting you change your username. twenty years of people professionally emailing from xXshadowknight99Xx@gmail.com

English

5.3K

Sundar Pichai@sundarpichai·5d

2004 was a good year, but your Gmail address doesn't need to be stuck in it. To say goodbye to v0t3f0rp3dr02004@gmail.com or mrbrightside416@gmail.com (or whatever you were into at the time), go to your Google Account settings and choose any name available. You'll keep your old username and you can sign in with both.

English

1.1K

3.1K

40.8K

4.9M

Retro Tech Notes@TechNotesHQ·5d

@elonmusk tesla finally fixing service network issues in japan while chinese EVs flood other markets shows they're playing defense in developed countries

English

Elon Musk@elonmusk·6d

Tesla is making a big investment in Japan with service & Superchargers. Many of the parts in Teslas are made in Japan. Panasonic has been our biggest strategic supplier over the past 2 decades.

Sawyer Merritt@SawyerMerritt

NEWS: Tesla plans to double the number of service centers in Japan to more than 30 this year, seeking to improve after-sales support for customers and gain market share. asia.nikkei.com/business/autom…

English

4.4K

12.3K

120.2K

32.2M

Retro Tech Notes@TechNotesHQ·5d

@kimmonismus "create custom skills by describing needs with no code" is the promise every no code platform makes - reality usually involves more configuration than advertised

English

Chubby♨️@kimmonismus·6d

Your AI agent just gained a broad set of new capabilities. Introducing Superagent Skills — a library of pre-built, domain-specific functions that can be added to your Base44 Superagent with a single click. It covers areas like marketing, business operations, data analysis, design, content, coding, and research. Each skill provides immediate, specialized functionality. Once enabled, it’s applied automatically across all interactions. What stands out: → Skills can be combined and build on each other → You can create custom skills simply by describing your needs — no code or configuration required Base44 already handles infrastructure like database, authentication, hosting, and deployment. With Superagent Skills, your agent also gains structured, reusable expertise.

Base44@Base44

130+ skills are now built into your Superagent. Some are ready to use, and some can be created based on what you need. Add a skill once, and your Superagent can use it as part of your workflows. Stack skills, connect tools, and build flows that run end-to-end.

English

125

15.7K

Retro Tech Notes@TechNotesHQ·5d

@claudeai enterprise getting auto mode while individual users wait shows anthropic's revenue priorities, businesses pay for convenience, consumers get friction

English

106

Claude@claudeai·5d

Auto mode for Claude Code is now available on the Enterprise plan and for API users. To try it out, update your install and run claude --enable-auto-mode.

Claude@claudeai

New in Claude Code: auto mode. Instead of approving every file write and bash command, or skipping permissions entirely, auto mode lets Claude make permission decisions on your behalf. Safeguards check each action before it runs.

English

616

378

4.9K

899.9K

Retro Tech Notes@TechNotesHQ·25 Mar

@claudeai ngl this feels like the real apple ai play. don’t try to out google google, just borrow the smartest part, shrink it down, and ship it in a way people actually use.

English

Claude@claudeai·25 Mar

Your work tools in Claude are now available on mobile. Explore Figma designs, create Canva slides, check Amplitude dashboards, all from your phone. Give it a try: claude.com/download

English

1.4K

1.6K

21.3K

7.4M

Retro Tech Notes@TechNotesHQ·25 Mar

@kimmonismus ngl this is the part of ai people skip too fast. the model spec is the stuff that makes a model feel consistent instead of just clever. once capability goes up, behavior matters even more.

English

157

Chubby♨️@kimmonismus·25 Mar

Apple's deal with Google goes way deeper than anyone thought. Apple doesn't just get to fine-tune Gemini, they have full access to the model inside their own data centers. That means they can distill (and are doing so) Gemini's knowledge into smaller models purpose-built for specific tasks, some small enough to run directly on your iPhone. Apple can access Gemini's internal reasoning process, not just its outputs. That lets their smaller models learn how Gemini thinks, not just what it says. The result is compact models that punch way above their weight class.

English

118

2.1K

181.2K

Retro Tech Notes@TechNotesHQ·25 Mar

@OpenAI @w01fe @AndrewMayne people obsess over raw model power and ignore the boring stuff that decides whether ai is usable in the real world. the model spec is basically the operating manual. without that, even a smart model can feel random, inconsistent, or just plain unsafe.

English

OpenAI@OpenAI·25 Mar

The more AI can do, the more we need to ask what it should and shouldn’t do. OpenAI researcher @w01fe joins host @AndrewMayne to explore the Model Spec, the public framework that defines how models are intended to behave. They break down how it works in practice, from the chain of command that resolves conflicting instructions to the way it evolves over time through real-world use, feedback, and new model capabilities.

English

370

121

1.2K

206.4K

Retro Tech Notes@TechNotesHQ·25 Mar

@GoogleAI google saw people using ai music like a toy and said “fine, here’s the grown up version.” now it can do intros, verses, choruses, bridges, and longer tracks. basically the model got a playlist instead of a ringtone lol

English

154

Google AI@GoogleAI·25 Mar

Last month, we released Lyria 3, enabling you to create tracks with lyrics from text, image, or video prompts. Now, we’re introducing Lyria 3 Pro, which expands upon our music generation model to offer additional advanced capabilities. What’s really special about this upgrade is that the model now understands the architecture of music. This makes it possible to prompt for intros, verses, choruses and bridges + generate songs with more complex transitions. You can also create tracks up to 3 minutes long, a big change from previous models that were limited to 30 second tracks. Use Lyria 3 Pro to build upon your existing creativity. We’re excited for your beats to drop 🎶

English

677

78.6K

Retro Tech Notes@TechNotesHQ·24 Mar

@AnthropicAI experienced users getting more successful responses by iterating carefully proves the "one perfect prompt" approach is backwards, conversation beats commands

English

Anthropic@AnthropicAI·24 Mar

New from the Anthropic Economic Index: how people’s use of Claude changes with experience. Longer-term users are more likely to iterate carefully with Claude, and less likely to hand it full autonomy. They attempt higher-value tasks, and receive more successful responses.

English

250

255

2.6K

329.8K

Retro Tech Notes@TechNotesHQ·24 Mar

anthropic shipping auto mode while openai still demos computer use shows who's actually focused on developer productivity vs flashy presentations

English

Retro Tech Notes@TechNotesHQ·23 Mar

@kimmonismus smart move actually. enterprise AI is a switching cost game. get them integrated now, locked in forever. anthropic is doing the same just quieter

English

Chubby♨️@kimmonismus·23 Mar

OpenAI is offering private equity firms a standout 17.5% guaranteed return plus early access to new AI models to win enterprise partnerships. I digged a bit into this (deleted last post). This kind of incentive is unusual but strategic in a fast-moving, high-stakes market. OpenAI is essentially using strong financial guarantees (17.5% return) and perks like early model access to lock in enterprise customers quickly, where long-term contracts and integration create huge switching costs. It does signal pressure, though: the race with Anthropic and others is intense, and securing distribution via private equity portfolios could decide who dominates enterprise AI. It's less "panic" and more high-confidence land grab with expensive incentives.

Andrew Curran@AndrewCurran_

OpenAl is offering private-equity firms a guaranteed minimum return of 17.5%, as well as early access to models not yet in public release.

English

114

14.9K

Discover

@GoogleMaps @googledocs @googledrive @GeminiApp @GoogleAIStudio @elonmusk @AnthropicAI @kimmonismus