Teachable Machine is a service that lets anyone train a neural network without coding ✍️

Teach a Computer to Recognize Images, Sounds, and Poses

👀 Want to understand how AI works? You should definitely try this tool! Google created an application that makes it easy to learn the basics of machine learning in a hands-on way.

All you need is a device with a working webcam and an object, gesture, or movement for the AI to recognize.

👀 The neural network learns from your examples, remembers your movements, and matches them with programmed responses — reacting with GIFs, sounds, or spoken messages.

@AximoLoot

BEN2AI — an AI tool that removes backgrounds from photos and videos in just a few clicks.

➡️ Upload your image or video.
➡️ Click “Remove Background.”
➡️ Wait a few seconds, and your result is ready.
➡️ No editing skills or complicated software required.

Perfect for product photos, social media content, thumbnails, and quick design tasks.

@AximoLoot

Amazon Shut Down an Internal Ranking That Tracked Employees’ AI Usage

🙂 Employees were competing to climb the leaderboard by burning through millions of AI tokens, often without considering whether their usage was creating any real value.

👀 The problem? It was reportedly costing the company millions of dollars in AI-related expenses.

@AximoLoot

The Amish Have Started Using ChatGPT for Work

The Amish are a religious community known for rejecting most modern technology. Yet, nearly 5% of Amish people in Holmes County, Ohio, are now using AI.

They have shared computers with internet access — strictly for business purposes and protected by heavy content filters. These computers are used to write letters, draft contracts, and manage records.

For farming, manufacturing, and business operations, some Amish have turned to ChatGPT. A few even use AI to write code and integrate it into their business tools.

The most interesting part? Phones.

🛑Smartphones are generally prohibited in Amish communities, and many Amish use only basic flip phones. But now, some of them are calling... ChatGPT.

Late last year, OpenAI launched a phone-based version of its assistant. By dialing +1-800-242-8478, users can speak with ChatGPT and get answers by voice. The feature never became particularly popular, and many people have never heard of it, but it has proven genuinely useful for the Amish.

🛑Those who use AI don't see it as a threat to their traditions. In fact, many believe it helps sustain their way of life.

“If we tried to operate using only fax machines and voicemail, I’d have to shut down,” says metalworking company owner Ian Wengerd.

@AximoLoot

Dreamina: CapCut is giving away AI image, video, and music generation for free

The teams behind CapCut and TikTok have launched Dreamina — an all-in-one AI creation platform with no subscription required.

What it can do:

🛑 Generate images with SeeDream 4.1
🛑 Create videos with SeeDance 2
🛑 Generate music directly inside the platform
🛑 AI-powered photo editing: fix faces, remove unwanted objects, expand backgrounds
🛑 An AI agent that can turn a simple idea like “I need a Reel cover about coffee” into a complete prompt and generate 4 versions in different styles

Pro tip from users: sign in through TikTok — it seems to come with fewer restrictions.

🛑At first glance, this looks like another serious competitor to Midjourney, especially for creators who don't want to pay a monthly subscription for images.

@AximoLoot

Bots Now Outnumber Humans on the Internet

👀 Bots have officially overtaken humans online. AI agents, crawlers, and automated systems now account for 57% of all internet traffic, while real people make up just 43%. It's the first time in history that human activity has become the minority.

👀 The numbers are making the "Dead Internet Theory" feel a lot less like a conspiracy and a lot more like reality. Cloudflare co-founder Matthew Prince previously suggested that AI would dominate web traffic by the end of 2026, but the shift appears to have happened much sooner than expected.

@AximoLoot

Grok Launches Imagine Agent Mode

🛑A new Imagine Agent Mode (Beta) has arrived in the web version of Grok, transforming the AI from a simple assistant into a full-fledged creative agent.

The system operates within a seamless, continuous workspace where it independently handles the entire creative process: planning ideas, generating content, editing, and repeatedly refining the final result.

🛑Users only need to describe the task — whether it's creating a short film, a manga, or marketing content — and the agent takes care of the execution from start to finish.

@AximoLoot

ChatGPT helped sell a house in 5 days without a real estate agent.

👀 A man from the US assigned the entire process to AI: from estimating the property value to communicating with buyers and preparing documents. In the end, the deal was closed in just 5 days.

👀 For comparison, the average time to sell a property with an agent is around 25 days, and the commission is usually 5–6% of the total price. In his case, that could have cost him up to $25,000.

A ChatGPT subscription cost him $20. Think about it 😉

@AximoLoot

AI detects ADHD in adults using medical records

Researchers in Sweden have developed an AI tool that can identify signs of ADHD in adults by analyzing medical history.

🛑The model was trained on records of nearly 8,000 patients and looks for hidden patterns, such as birth complications or substance-related issues that may be linked to undiagnosed ADHD.

🛑In tests, AI correctly identified 80% of people with ADHD and ruled it out in 77% of those without the condition.

However, it still struggles to distinguish ADHD from depression and anxiety because the symptoms often overlap.

@AximoLoot

Does Your Idea Have Million-Dollar Potential — or Is It Bound to Fail?

🗄 Greg Brockman shared a powerful approach for Codex that helps break down a startup idea from every angle.

🗄 It’s simple: describe your idea, add the context — and AI starts analyzing it from all sides: the market, demand, competitors, strengths, and weaknesses.
Basically, it’s a fast idea audit before you spend your time and money on it.

You can start with a single prompt:

npx --yes codex-startup-pressure-test-skill@latest

@AximoLoot

AI Has Created a New Type of Escort

👀 Forbes writes about women who specialize in clients from the AI and big tech industries. Their main advantage? The ability to spend hours discussing neural networks, crypto, biohacking, and the future of humanity.

✍️ One woman featured in the article charges $3,500 per hour, another $5,000. Some say their clients often include Nvidia employees, startup founders, and people who became wealthy during the AI boom.

@AximoLoot

Cheap Talking Avatars from PrunaAI

Upload a face photo + text or audio — and get a video where that person speaks. The p-video-avatar model has just launched on Replicate.

✍️Features:

• 30 voices included, 10 languages available
• Add text or your own audio — lip sync is generated automatically
• Control the speaking style with prompts: “speak excitedly”, “like a narrator”, etc.
• Add a separate description of what happens in the scene


The main advantage is the price: $0.025 per second in 720p. A 10-second video costs just 25 cents. 1080p costs twice as much.

ℹ️Possible uses: talking avatars for Reels, translating the same script into different languages with the same face, or animating a podcast host portrait with ready-made audio.

@AximoLoot

AI 3D Generators vs Real 3D Printers: Who Wins?

AI promises: “Give me an image — I’ll give you a printable 3D model.”

Reality? Broken meshes, flipped normals, messy geometry, and slicers struggling with random floating parts.

➡️ A Habr author tested Tripo, Meshy, 3DAIStudio, and HiTech 3D on free plans. They fed them a T-Rex image and waited for a miracle.

The result? More like a “39°C fever dream.”

➡️ The conclusion: Image-to-3D is great for concepts and rough drafts, but not ready-to-print FDM models yet. If you want to actually print something — Blender and manual work are still part of the process.

@AximoLoot

AI File Sorter — a new AI-powered file manager that replaces the default one in Windows 11.

🗄Fast search by content and metadata

🗄Bulk document renaming

🗄Automatic file organization into folders

🗄Preview changes before applying them

🗄Supports Windows, Linux, and macOS

🗄Works via API or locally

@AximoLoot

Found a cool screen recorder that saves tons of time

You know the struggle: you record your screen, then spend hours editing — zooming into important parts, cutting pauses, cleaning up audio, adding subtitles...

FocuSee does it all automatically. Here’s what it can do:
🎮 Automatic zoom & cursor tracking — it highlights important areas when you click.
🎮 AI editing in minutes — removes pauses and filler words (like “uhhh”) from audio.
🎮 Auto subtitles in 50+ languages
🎮 Voice enhancement — removes background noise and breathing sounds.
🎮 3D Motion & annotations — adds smooth cursor movements and highlights actions.
🎮 Remove/replace webcam background
🎮 AI avatars
🎮 Mobile screen recording

The biggest advantage: you get a professional-looking video within minutes, without spending hours editing.

@AximoLoot

Clone any website in a few clicks: grab a service that turns web resources into ready-to-use prompts for AI models.

It works incredibly simply:

📼 Drop a link — the service scans the page and extracts everything valuable: colors, fonts, icons, animations, page structure, UI features, and site logic;

📼 Then it creates a complete dossier of the website along with a screenshot;

📼 The generated file can be fed to an AI and used to get a highly similar result;

📼 You can use it as a reference for your own project or quickly transfer the solutions you like;

📼 The source code is open, and you can run the tool locally on your own computer.

@AximoLoot