Dare Divil Collective

Welcome to the collective!

recent posts

about

  • ChatGPT Apps Beta

    5 Apps That Might Change How We Use AI

    So ChatGPT quietly rolled out what some people are calling an “Apps beta.”

    It’s not officially an app store yet, but it’s definitely starting to feel like one.

    The idea is simple. Instead of jumping between apps, you stay inside ChatGPT and connect services directly into the conversation. You can call them up using the at command, and ChatGPT can even suggest apps based on what you’re doing.

    Here are five apps that already stand out to me.

    Spotify and Apple Music let you describe a mood or activity and instantly play the right music without leaving the chat. DoorDash and Instacart take it a step further. If you’re talking recipes, ChatGPT can help you order ingredients on the spot. TripAdvisor ties into travel planning, letting you search and book while you’re already asking questions.

    What’s interesting isn’t just the apps themselves. It’s the direction this points to. If developers can build directly inside ChatGPT, this starts looking less like a chatbot and more like a platform.

    I’m curious how far this goes.

    Is this just a convenience feature, or are we looking at an early version of something that could eventually compete with traditional app stores?

    Drop a comment and let me know what you think, and which app you’d actually use.

  • PAN: The New “World Model” That Actually Remembers

    While Microsoft was shaking up local AI agents, MBZUAI introduced something completely different: a world model called PAN. Instead of generating single video clips that instantly forget everything, PAN builds and maintains a continuous digital world that updates every time you give it a new instruction.

    Most video models wipe their memory the moment a clip ends. PAN doesn’t. If you tell it to turn left, then speed up, then pick up a block, each action continues from the last. It’s not producing disconnected visuals—it’s simulating cause and effect. That’s why researchers call it a world model rather than a video generator.

    The model uses Qwen2.5-VL for reasoning and a video generator adapted from Wan 2.1. Instead of letting visuals destroy consistency, PAN keeps reasoning in a stable internal space and then translates those states into video. That prevents objects from morphing or drifting during long sequences, something most video models completely fail at.

    A big part of PAN’s stability comes from its causal, chunk-based refinement system. The model only looks at past frames—not future ones—which forces it to respect continuity. They even add slight noise to prevent it from over-focusing on tiny pixel details and losing track of the scene’s big picture.

    Training this model was a massive project. MBZUAI used 960 Nvidia H200 GPUs, recaptioned thousands of videos to emphasize motion and cause-and-effect, and filtered out anything that wouldn’t help with real-world simulation. The payoff is huge: PAN scores 58.6% overall on action-simulation tasks, outperforming many commercial world models that avoid publishing their numbers.

    PAN even works as a planning tool. Plugged into an O3-style reasoning loop, it hits 56.1% accuracy on simulation tasks, making it strong enough to act as the “what happens if I do this?” module inside future AI agents.

    This is the direction the industry is moving toward—AI systems that can understand actions, predict consequences, and maintain stable worlds over time. PAN is one of the first open-source models to make that idea feel real.

  • Apple iOS 26.2 Explained — Executive Shakeups, AI Shifts, and What This Means for Users

    Apple just released iOS 26.2, and while the update brings performance fixes and new features, the bigger story may be what’s happening behind the scenes. Several high-level Apple executives are leaving or transitioning roles, raising questions about Apple’s direction in AI, design, and long-term strategy.

    In this video, I break down what iOS 26.2 actually improves, why this update feels like a course correction, and how recent leadership changes could impact Apple’s future products and software. From AI strategy shifts to design changes and system stability, this is a grounded look at what matters — without the hype.

    If you care about Apple, accessibility, and how these changes affect real users, this one’s worth watching.

    Click here to watch the video

  • Holiday Shopping While Blind — Decorations, Crowded Stores, and How I Really Navigate

    Holiday shopping looks very different when you’re blind — decorations move, mall layouts change, and walkways aren’t always where you expect them to be. In this video, I share how I actually navigate crowded stores using my cane, sound, and mental mapping, and why simple descriptions matter more than people realize. Watch the full video here and let me know what the holidays are like where you shop.

    Click here to watch video

  • Ray-Ban Meta 20.0 Update: What It Means for Blind Creators Using Gen 1 Glasses

    The new Ray-Ban Meta 20.0 update brought more changes than the official release notes admitted, and as someone who uses these glasses every day as a blind creator, I wanted to break down what actually matters. This is especially important for anyone still on the Gen 1 model, because this update quietly showed just how far Meta is pushing users toward Gen 2.

    The biggest improvement for me is that slow motion and hyperlapse finally work on both Gen 1 and Gen 2. That matters because a feature like hyperlapse lets me walk around, capture the moment, and still keep listening to music or an audiobook since hyperlapse does not record audio. For workflow, that means I can stay focused and stay in motion without losing my rhythm. Slow motion being available on Gen 1 also brings new creative possibilities without upgrading hardware.

    But we have to be honest. Gen 1 users are starting to get left behind. The five minute recording limit still stays exclusive to Gen 2, and that affects creators who record longer clips or rely on the glasses for hands-free filming. When you create content the way I do, every limitation shows up in your workflow fast. This update also reminds us that Meta now has to support multiple devices at once, and Gen 1 is slowly getting less attention.

    The new app connections layout inside the Meta app is a small but useful improvement. As a blind creator, clarity and organization matter because I navigate everything through a screen reader. Having connected and non-connected apps separated, and seeing what each app is paired with, helps me stay in control of my setup. It also removes the frustration of hunting through settings to disconnect or re-connect something.

    The biggest disappointment is still the missing Conversation Focus mode. That feature was announced months ago, and it was supposed to help isolate the voice in front of you and reduce background noise. Features like that are not just upgrades for blind creators. They’re essential tools for independence. Whether you’re at an event, in a loud venue, or recording something on the go, having clearer audio could change the entire experience. The fact that this still has not arrived on either generation says a lot about how slow some of the accessibility-friendly features are rolling out.

    Overall, the 20.0 update gives Gen 1 owners a few exciting tools, but it also makes it clear that Meta is steering creators toward the newer hardware. As someone preparing courses for blind users and creators, updates like this are a reminder that we need to track features version by version, not assume both models will grow evenly. The good news is that hyperlapse and slow motion expand what’s possible on Gen 1. The bad news is that the gap between Gen 1 and Gen 2 is becoming more obvious every month.

  • How to help navigate crowded Spaces For a blind individual. 

    Have you ever walked through a loud mall or packed store during the holidays and thought nothing of it? For someone who’s blind, those spaces sound completely different — and navigating them takes a whole mix of listening, patience, and awareness most people never think about.

    In today’s Unseen Adventures video, I break down how I navigate with echolocation, what throws me off, and what actually helps.

    If you’ve ever wondered how blind people move through the world, this one’s for you. Click here to watch

  • People always ask why my white kane is long and here is why?!

    People always ask me why I use such a long cane — so I finally made a video breaking it down. I’m 5’7, but my cane is 58 inches… and there’s a real reason for that. Reaction time, safety, confidence — it all plays a part.

    I even went outside and showed how I use it in real streets so people can see the difference.

    Click here to watch

  • Holiday Shopping rush and blindness

    Holiday shopping is stressful for everybody… but when you’re blind, it hits different.

    Checkout lines twisting everywhere, carts blocking the aisles, people saying “over there” like that’s a real direction — it gets wild out here. And somehow, people still go silent when you’re about to bump into them.

    I made a whole video about what it’s really like during the holidays. If you want to understand what we go through (and maybe laugh a little), watch it here 👇Click here

  • Today we look at the application Scribe Me: is this one of the new top AI applications for Accessibility. 

    Today I checked out an app that honestly surprised me. It’s called Scribe Me, and it’s built from the ground up for visually impaired folks — PDF reading, PowerPoints, image descriptions, Live Assist… all in one place.

    I tried it on real files, and the results were way better than I expected. Even with my voice still sounding a little raspy, this was worth filming.

    If you want to see how it works and why it could become one of the best AI tools for blind users, check out the video here 👇Scribe Me App Review: The Best AI Scanner & Live Assist Tool for the Blind?

  • Microsoft just dropped something that honestly feels like a shift in the AI world 

    FARA-7B, a compact computer-use model that’s powerful enough to run locally without burning your machine to the ground. This thing isn’t another bloated agent chained to the cloud. It’s built to run on regular hardware and still handle real tasks. 

    What makes it stand out is its simplicity. Most agents use a giant stack of subsystems that click, scroll, guess at the screen, and call multiple helper models behind the scenes. FARA-7B does the opposite. It looks directly at a screenshot and decides what to do next. No scaffolding. No accessibility-tree parsing. No five-model circus happening backstage. Just one model handling everything. 

    The magic comes from Microsoft’s synthetic data engine, Faragen. Instead of harvesting human browsing logs, they trained the model with AIs performing tasks across more than 70,000 websites. These weren’t perfect robotic demos either—they included mistakes, retries, scrolling, searching, and all the messy behavior humans actually do. After that, three separate AI judges verified every session to make sure the actions matched the on-screen reality. 

    All of that added up to over a million individual actions used for training, giving the model extremely grounded behavior. The final result is an agent that doesn’t hallucinate clicks and doesn’t go rogue because it learned from full sequences of real web interactions. Most importantly, it runs locally, so latency drops and privacy shoots way up. 

    Performance-wise, the numbers are wild for a 7B model. On benchmarks like WebVoyager, WebtailBench, and DeepShop, FARA-7B matches or beats much larger agents while using a fraction of the tokens. A full task costs about two and a half cents compared to thirty cents for big GPT-powered agents. That’s a massive difference in both speed and affordability. 

    This is exactly what people hoped AI agents would eventually become—small, private, cheap, and accurate. FARA-7B is one of the first real signs that computer-use models are moving away from cloud-