βRealtor.com, Homes.com, Redf Fin, YouTube LinkedIn Instagram sometimes Facebook, local websites.β
What the smart people are recommending. 7867 books, tools, and products endorsed by the thinkers absorb.md tracks. Ranked by how many times each has been recommended across compiled podcasts, papers, posts, and tweets.
βRealtor.com, Homes.com, Redf Fin, YouTube LinkedIn Instagram sometimes Facebook, local websites.β
βWe're super excited about the roll out of our platform is with Sirino. And so that is going to be heavily marketed, but within Sirino, you're going to be able to organize all of your ecosystem data, lβ¦β
β\texttt{DustPy}, which simulates the transport and evolution of dust particles including detailed coagulation and fragmentation.β
βOur findings surface significant tensions between the relationality of our participants' queer art practices and the perceived anti-relationality of GenAI development and use. We detail how our particβ¦β
β# Release of 3 Flash replacing the current version of 2.5 Flashβ
βAnnouncement of general availability for 2.5 Proβ
βAnnouncement of general availability for 2.5 Pro and Flashβ
βintroduction of Gemini 2.5 Flash-Liteβ
βIf anyone wants to go and read more about many of the things we've talked about today, where do they go? Substack. I'll link below.β
βwhat they've been building is this new tool called Stanley that uses AI, looks at your feed, looks at your tone of voice...and tells you what you should post, makes those posts for you.β
βWolf bench is like a really great tool... we've been pushing Wolf Bench here to everyone. Peter Steinberger heard about Wolf Bench.β
βthere are some fixes going on with Lama CPP with the vision capabilities and so on. But yeah, it's going well.β
βMaking sure you're using encrypted messaging, understanding the internet of things, and just how bad things can be on there, how insecure.β
βTry SAM 3 in the Segment Anything playground.β
βTry SAM 3 in the Segment Anything playground.β
βIf you're locked out of OpenClaw right now, then Kilo Claw is hands down the fastest and most secure way to get your OpenC environment back up and running.β
βI'm an investor in a company called Stanto...they've been building is this new tool called Stanley that uses AI, looks at your feed, looks at your tone of voice...and tells you what you should postβ
βAnd my YouTube channel is just my name. We'll try and collab with you on this video.β
βPlay with these toys called AI tools. Someone has given you a free iPhone which you can just use called AI tools on the free account.β
βSUM, formerly Cryptotax Calculator, was built specifically to solve this problem. It supports Turboax along with over 3,500 exchange wallet and crypto integrations.β
βSo on x cryptohaze on substack cryptohaze as well. I write my monthly bionthonly newsletter and yeah you'll hear me on programs like yourselves.β
βWe present gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models that push the frontier of accuracy and inference cost.β
βIf you are on one of those paid plans up under the selector here, you should see Thinking 5.4 and Pro 5.4. These are the brand new models that you can use.β
βI actually built a little app for myself...I actually use this flash light model for that because it's really fast...But also it's really cheap to use.β
βHonestly, this is one of my favorite features Notebook LM has actually added. Like, look at these animations.β
βIt's a new speech recognition model achieving best-in-class accuracy across 25 languages, which is purpose-built for transcription workloads.β
βPut Claude to work on your computer with this feature...Claude can use your computer to point click and complete tasks.β
βI always thought that Google's live features are super underrated and not talked about enough because you could literally get it to like teach you how to do things, just show your screen and have it wβ¦β
βPerplexity computer loads tax modules for US federal tax questions. computer can help with hard parts of tax work that usually cost people the most time and money.β
βThey also released Lia 3 Pro, which is kind of like Google's version of Sunno, where you can have it generate any music you want.β
βPersonally, I'm a big fan of Slack. I do all of my work inside of Slack. Even my open claw is built for me to talk to it through Slack. So, this is something I'll definitely be playing with.β
βif you've never played with like give it a shot it it's kind of a blastβ
βif you're building an app that needs a voice agent that sounds pretty much as good, if not a little bit better than 11 Labs, maybe check out Smallest.ai.β
βMistrol also released a texttospech model this week and this one is an open weights model that you can actually run locally on your own computer.β
βThis company, Love Art AI, released a new feature called move object...You can actually play around with this one for free as well.β
βopen Da Vinci Resolve and show me where the magic mask feature isβ
βIt looks like you're using OBS Studio, which is a popular program for streaming and recording.β
βAdobe announced they had 5 billion of quote AI influenced revenue this year... to me, it don't count unless it's net new bookings... Adobe gets the worst score.β
βThis feature is available on all plan types. You should even be able to use it if you're using Claude for free.β
βinteractive learning is rolling out starting today to all logged in ChatGpt users... if you need help with math or sciency topics, there is a bunch of cool built-in visuals here.β
βI just think it's really really cool. You can generate any image you want and then go and recompose it... from what I understand, it's available in all plans on Canva.β
βthey've just released their eighth version of MidJourney... MidJourney V8 is much better at following detailed directions, has a much better ability to understand your aestheticsβ
βI already think it's more impressive than midjourney if I'm going to be honest... it nailed it. It did the exact thing it was supposed to.β
βIf you haven't already, check out stitch.withgoogle.com. I'm really quite impressed because now what we can do is we could literally take this design and throw it into codeex or claude code or cursorβ
βAndre Carpathy... recently released this US job market visualizer to show which jobs are most at risk of AI sort of taking your job... If you want to click around in here, of course, I'll link this upβ
βThere's a new feature inside of Adobe Photoshop. Now, inside of Photoshop, you can actually give it text prompts to adjust images... it's available on web and mobile.β
βfrom everything I hear, it's a really impressive model, especially for an openweight model... it is available now. You can use it inside of Nvidia's platform, Perplexity, Open Router, and Hugging Faceβ¦β
βif you're a developer and you're building something... This model could be a model that you build with that can do that kind of thing if that makes sense.β
βChatgPT for Excel came out this week... it can read your data and it can also add data to the sheet... teams can describe what they need in plain language and chat GPT will create or update live Excelβ¦β
βthey launched scheduled tasks in Claude Code... There's also a new code review feature inside of Cloud Code... Apparently, Anthropic uses this code review internally.β
βGoogle rolled out a new update to maps. You can use it to ask maps questions and have conversations about your surroundings.β
βGoogle also rolled out some new Gemini features in Docs, Sheets, Slides, and Drive... go to a first draft instantly. You could polish things. You could match the voice or writing style.β
βBox should win. Box should reacelerate to 20 or 30%... It has all the AI and it has the data. It has trillions of documents, right, that you can't even find. Now I can talk to my documents.β
βThey bought Moveworks just now and they said, you know, hey, we have we had an AI story before we bought a small company clearly getting some traction... I thought it was kind of a shrewd move.β
βI worked at Survey Monkey for a couple years and this was one of the things that we did was like to make sure the voice of the customer was sort of both programmatically pulled into the companyβ
βApple is partnering with Google Gemini right smart strategy in your opinion? I think it's the only strategy right now for them.β
βIf you open up Siri now, you can ask either Chat GBT or Siri a question.β
βYou see what Microsoft does with Windows and it feels a little like they're cramming too much AI in your face, you know, with co-pilot there.β
βWe propose T-Learners layered into the G-Formula for this setting, building on literature from causal machine learning and identification in sequential settings.β
βMeta's new Muse Spark model is getting lavished with praise by Wall Street...this model is still below the capability of chat GPT and Claude...maybe Wall Street needs to chill out on this one.β
βyou can go to rw.xyz XYZ. You can see there's six different versions of tokenized coin which none of them is actually equity on Coinbase.β
βThere's this webpage called the Risk List, uh it's uh spelled uh with a Q instead of a C, uh by this company called Project 11, where they have this uh you know, this dashboard that gives you a live vβ¦β
βYeah Quen 36 looks like a really good model. I haven't had a chance to play around with it yet. I definitely will.β
βSAM audio is a cutting-edge audio model with applications for musicians, audio engineers, video creators, and tinkerers of all skill levels. Learn more about SAM Audio today.β
βSteve Newman, who's been doing some great writing about AI on his Substack newsletter, gives a good example of this in the context of his being disappointed over GPT5.β
βIf you want to understand mythos, you need to understand how Anthropic actually organizes their models. So at the bottom you've got Haiku, which is cheap and lightweight. Then you've got Sonic, which β¦β
βIf your Android phone has 8 GB of RAM or more and was released in the last few years, you can probably run the Gemma 4 models. If it has 12 GB of RAM, you can probably handle the models too.β
βAnd of course, the smaller the model is, you can see down here you've got Gemma 3 1 billion parameters that is going to be a model that is going to be easily run on the older devices that we mentionedβ¦β
βSomething slightly more accessible, which you might have heard of before, is the Ising model. This is a model of a magnet which consists of a bunch of tiny magnetic elements arranged in a grid, each oβ¦β
βThis is called the XY model. If you are curious and want to learn more, all of these models can also be considered in three dimensions, or even higher, or on other strange non euclidean geometries.β
βIf you are using certain open- source models, maybe they may not have the same performance as Opus 4.5 or GPT 5.4 or GPT 5.3 codecs. And you should take that into account because remember every singleβ¦β
βWe present gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models that push the frontier of accuracy and inference cost.β
βWe propose CRISP, which instead jointly ranks all cited papers within a citing paper using large language models (LLMs).β
βAccess to my Content Catalyst custom-built AI tool that help you write content that actually sells.β
βyou could save a lot of money by working with GPT 40 mini instead of GPT 40β
βQuick start: the Iris dataset in Keras and scikit-learnβ
βWe also include results for Command R7B which shares capability and architectural similarities to Command A.β
βfor um try uploading stuff uploading files it's pretty phenomenal if you've got a really long you know paper to readβ
βimages in images out um ... there's a lot of cool stuff you're ricking on a birthday card you're probably going to want to use both um make a little Custom Imageβ
βyou can go online test it for yourself the the web address is thereβ
βwe think we got there earlier this year um with the o1 releaseβ
βtell me what's your favorite open AI API I think the new realtime API is pretty awesomeβ
βWhen these EMVS methods are compared to Stochastic Search Variable Selection (SSVS), the EMVS methods surpass SSVS both in terms of computational speed and correctly identifying significant variables.β
βTo preserve the computational speed of EMVS we also implemented the Stochastic Dual Coordinate Descent (SDCA) algorithm.β
βI'd recommend using neotron 340b to enhance and make it better and we've already seen made llama better made all the other models betterβ
βyou can see with music, you can see you know uh Suno and IDO and so forth.β
βat the moment I've got vs code with co-pilot and I've got cursor and I've got Zed I'm still sticking in vs code co-pilot is the one that I use every dayβ
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βIf you want access to my AI tool stack - Dovetail, Arize, Linear, Descript, Reforge Build, DeepSky, Relay.app, Magic Patterns, Speechify, and Mobbin - grab Aakashβs bundle .β
βAdapting foundation models under resource budgets relies heavily on Parameter-Efficient Fine-Tuning (PEFT), with LoRA being a standard modular solution.β
βUsing pre-trained word embeddings in a Keras modelβ
βAndrej Karpathy confirmed it. The CLIs to install today: GitHub CLI ( gh ) - Claude is aggressively competent at it.β
βFirecrawl CLI - Clean web scraping without MCP overhead.β
βThe make slides skill from the episode includes Puppeteer so Claude can screenshot its HTML output, measure text overflow, and fix issues automatically.β
βThe web research skill replaces Claudeβs unreliable default web search (basically Google first-page results) with Tavily for higher-quality search and Firecrawl for clean markdown scraping.β
βTell Claude βuse your ask user questions toolβ and it generates a custom UI with checkboxes and input fields right inside the terminal.β