Word of Lore.ai
LightNews LightNews
banner
wordoflore.bsky.social
Word of Lore.ai
@wordoflore.bsky.social
5 followers 2 following 180 posts
In the space between AI's promise and its practice, we find the stories that matter. Practical insights. Human purpose.
Posts Replies Media Videos
wordoflore.bsky.social
From "this will replace all creatives" to "it's a very expensive sketching tool" in record time — the lifecycle of every AI product announcement.

#BuildAI

wordoflore.ai/veo-3-1-nati...
Veo 3.1: native audio and reference controls
Veo is Google's latest attempt to teach computers how to make videos from scratch. Now in version 3.1, it's available for anyone willing to pay for early access, either through Google AI Studio or Ver...
wordoflore.ai
November 12, 2025 at 2:20 PM
wordoflore.bsky.social
The modern productivity calculation: will I trade my entire digital life setup for slightly less tab-switching? Apparently yes.

#ApplyAI

wordoflore.ai/openai-atlas...
OpenAI: Atlas browser
Imagine if your web browser and ChatGPT were the same thing. That’s what OpenAI has done with ChatGPT Atlas. Instead of jumping back and forth between tabs, you just talk to the AI right where you’re ...
wordoflore.ai
November 11, 2025 at 1:31 PM
wordoflore.bsky.social
We've reached the point where 'made by AI' and 'magazine quality' stopped being contradictions

#ApplyAI

wordoflore.ai/adobe-firefl...
Adobe Firefly: AI soundtracks and voiceovers
Adobe Firefly is a creative playground powered by AI, where you can make and edit images, videos, and even audio—all right in your browser. So, what’s new? Now, with just a click, you can create a so...
wordoflore.ai
November 10, 2025 at 1:37 PM
wordoflore.bsky.social
We've moved from testing if AI gets the right answer to testing if it reasons correctly—which is either progress or peak parenting anxiety for your code.

#BuildAI

wordoflore.ai/pydantic-eva...
Pydantic: Evals
Pydantic Evals is a tool for Python that lets you watch, step by step, how your AI agents go about solving problems. It’s made by the same people who built the popular Pydantic data validation library...
wordoflore.ai
November 7, 2025 at 1:32 PM
wordoflore.bsky.social
Discovering your AI is technically helpful but functionally useless—a special kind of product failure that only shows up when someone actually tries to use the thing

#BuildAI

wordoflore.ai/langsmith-mu...
LangSmith: multi-turn evaluation
Imagine you’re chatting with an AI, asking it to help you book a flight. It might give you the right answer to every single question you ask, but somehow, you still end up without a ticket. That’s whe...
wordoflore.ai
November 5, 2025 at 1:31 PM
wordoflore.bsky.social
A hundred dollars to watch a computer figure out the alphabet—the tuition for machine kindergarten is honestly more reasonable than I expected

#BuildAI

wordoflore.ai/karpathy-nan...
Karpathy: nanochat
nanochat is Karpathy’s attempt to strip LLM training down to its bare essentials. It’s about 8,000 lines of code, and it’s designed to be read and understood, not just run. Unlike the big, complicated...
wordoflore.ai
November 3, 2025 at 1:30 PM
wordoflore.bsky.social
Biology is now just another domain for natural language processing—we've successfully reduced cellular complexity to a tokenization problem

#BuildAI

wordoflore.ai/google-resea...
Google Research: the language of biology
Imagine you could talk to cells and ask them what they’re up to. That’s more or less what Cell2Sentence-Scale (C2S-Scale) lets you do. Built by Google Research and Yale, it’s an open-source model that...
wordoflore.ai
October 30, 2025 at 12:44 PM
wordoflore.bsky.social
Adobe's main selling point: we probably won't get you sued. Hell of a value proposition

#ApplyAI

wordoflore.ai/adobe-firefl...
Adobe Firefly: commercial-safe AI
Adobe Firefly is Adobe’s own set of AI tools, built right into the Creative Cloud apps you probably already know—Photoshop, Illustrator, and Adobe Express. There’s even a mobile app for your phone. Wh...
wordoflore.ai
October 29, 2025 at 12:37 PM
wordoflore.bsky.social
AI now turns napkin sketches into diagrams. We've finally automated the 'who's going to make a proper version of this?' question that ends every brainstorming session.

#ApplyAI

wordoflore.ai/chatgpt-figj...
ChatGPT: FigJam diagram creation
Imagine you’re chatting with ChatGPT, scribbling ideas on a napkin, or digging through old documents. Now, with the Figma app, all of that can be turned into editable diagrams you can actually use. H...
wordoflore.ai
October 27, 2025 at 1:08 PM
wordoflore.bsky.social
When 'not needing to buy a new computer' counts as a feature announcement, the bar has traveled to interesting places.

#ApplyAI

wordoflore.ai/windows-copi...
Windows Copilot: voice-first AI
Copilot is Microsoft's built-in AI assistant for Windows 11. Until now, some of its best features, like talking to it with your voice or letting it see what's on your screen, only worked on fancy new ...
wordoflore.ai
October 25, 2025 at 3:39 PM
wordoflore.bsky.social
'Download and run locally' went from the entire selling point to nostalgic tech memory in about five years flat.

#UnderstandAI

wordoflore.ai/the-desktop-...
The desktop AI gap
Imagine you’re a developer, and you want to try out the latest AI model on your laptop. Not so long ago, you could just download it and get to work. But now, these models have grown so huge and comple...
wordoflore.ai
October 24, 2025 at 12:49 PM
wordoflore.bsky.social
Everything requires its own API endpoint now. Still waiting for the `/v1/existential-crisis` endpoint for when the system becomes self-aware.

#BuildAI

wordoflore.ai/anthropic-ag...
Anthropic: Agent Skills
Agent Skills is a new way for Claude to learn new tricks. Imagine you could hand Claude a folder full of instructions, code, and resources, and Claude would know exactly when to use them. That’s what ...
wordoflore.ai
October 23, 2025 at 12:53 PM
wordoflore.bsky.social
We've discovered that understanding something and controlling it are unrelated skills. Every parent could have told AI researchers this.

#BuildAI

wordoflore.ai/sae-steering...
SAE steering: Delta Token Confidence
Imagine you’re trying to understand what’s going on inside an AI’s mind. Sparse Autoencoders, or SAEs, are a tool that lets us break down the AI’s thoughts into features we can actually make sense of....
wordoflore.ai
October 22, 2025 at 12:32 PM
wordoflore.bsky.social
We trust it to navigate the internet but not to spend money. That tells you everything about what the internet has become

#BuildAI

wordoflore.ai/gemini-compu...
Gemini: computer use
Google DeepMind has just released something called the Gemini 2.5 Computer Use model. In plain English, it’s an AI that can use computers almost like a person does. It can click buttons, type into for...
wordoflore.ai
October 21, 2025 at 8:40 PM
wordoflore.bsky.social
We've reached peak meta: needing AI therapists to explain what our AI is thinking. The trust issues are mutual.

#BuildAI

wordoflore.ai/fraunhofer-c...
Fraunhofer: circuit-based interpretability
Mechanistic interpretability is all about peering inside a neural network’s mind and asking, ‘How does this thing actually think?’ The usual way is to throw billions of words at the model and then ask...
wordoflore.ai
October 21, 2025 at 12:43 PM
wordoflore.bsky.social
We don't let doctors or pilots grade their own exams, but we did let the AI companies do it because they promised they'd be super honest about it.

#UnderstandAI

wordoflore.ai/the-benchmar...
The benchmark contamination problem
Here’s the problem: Nearly half of the questions on these AI tests are already in the model’s memory banks. Imagine if you sat an exam and the teacher handed you last year’s answer sheet. That’s what’...
wordoflore.ai
October 20, 2025 at 12:34 PM
wordoflore.bsky.social
We've graduated from autocomplete to autocorrect your entire value system. Progress!

#UnderstandAI

wordoflore.ai/uc-santa-cru...
UC Santa Cruz: AI shifts human meanings
Symbolic Interactionism is a fancy way of saying that we make up meanings together, through our conversations and interactions. The researchers at UC Santa Cruz wanted to see if AI could join in on th...
wordoflore.ai
October 19, 2025 at 8:17 PM
wordoflore.bsky.social
The academic equivalent of judging books by their AI-generated video covers—this would horrify scholars from literally any previous generation

wordoflore.ai/notebooklm-v...

#ApplyAI
NotebookLM: video overviews
Imagine you’ve got a pile of documents and you wish someone could just explain them to you. That’s where NotebookLM comes in. It’s Google’s AI tool that takes your documents and, with a feature called...
wordoflore.ai
October 17, 2025 at 7:25 PM
wordoflore.bsky.social
AI memory tool: every conversation is no longer a first date where you explain your job for the tenth time

wordoflore.ai/claude-memor...

#BuildAI
Claude: memory tool
Imagine you’re building something with Claude, and you want it to remember things from one conversation to the next. Anthropic has just released a memory tool for their Claude API that lets you do exa...
wordoflore.ai
October 17, 2025 at 12:45 PM
wordoflore.bsky.social
The eternal software cycle: create fifteen tools that don't talk to each other, then create tool sixteen to make them talk

wordoflore.ai/notion-ai-ag...

#ApplyAI
Notion: AI Agent
Notion is one of those tools that tries to be your digital brain—a place to stash your notes, documents, and databases. But now, they've given their AI assistant a serious upgrade. Meet the Notion Age...
wordoflore.ai
October 16, 2025 at 12:38 PM
wordoflore.bsky.social
Harvard's releasing a textbook on ML systems that 'actually work in the real world'—the fact that this is a selling point tells you everything about how we've been teaching this.

wordoflore.ai/harvard-ml-s...

#BuildAI
Harvard: ML systems textbook
Imagine you want to build machine learning systems that actually work in the real world, not just in a classroom. Harvard has put together a textbook for exactly that. It’s based on their CS249r cours...
wordoflore.ai
October 15, 2025 at 2:25 PM
wordoflore.bsky.social
Nothing says innovation like rediscovering that computers can actually compute things locally

wordoflore.ai/windows-ml-o...

#BuildAI
Windows ML: on-device inference
Windows ML is a tool from Microsoft that lets developers run AI models right on your own computer. Instead of sending your data off to some distant server, everything happens right there on your devic...
wordoflore.ai
October 15, 2025 at 12:32 PM
wordoflore.bsky.social
Parents can't read teen conversations unless the AI thinks they should. So we've outsourced 'should I tell mom' to the same system that writes their essays

wordoflore.ai/chatgpt-pare...

#ApplyAI
ChatGPT: parental controls
ChatGPT is that AI tool you might use to help with homework, write an essay, or just ask a weird question at 2am. Now, with parental controls, parents and guardians can set some ground rules for how t...
wordoflore.ai
October 14, 2025 at 9:53 PM
wordoflore.bsky.social
Google finally taught smart homes that humans don't speak in rigid command syntax. Only took a decade

wordoflore.ai/google-gemin...

#ApplyAI
Google: Gemini for home
Gemini for Home is Google's new AI that takes over from Google Assistant on your smart displays and speakers. It also upgrades your cameras, doorbells, and the Google Home app. Basically, if you have ...
wordoflore.ai
October 14, 2025 at 4:07 PM
wordoflore.bsky.social
We made shopping so terrible with technology that talking like an actual human being now counts as innovation

wordoflore.ai/google-searc...

#ApplyAI
Google Search: conversational shopping
Imagine searching for something online and just chatting with Google like you would with a friend. That's what AI Mode is: a way to ask questions in your own words and get answers, complete with pictu...
wordoflore.ai
October 14, 2025 at 1:25 PM