Qwen3.8-Flash is now free in Qoder through Sept 30 👀 Even free accounts can use it — no claim or paid plan required. A solid chance to try the model directly inside Qoder while the free access lasts. qoder.com/qoder

I wanted to see if Pexo could turn a complex company story into a fast-paced motion graphics explainer. So I gave it a clear direction: a 30-second vertical video about Sequoia Capital, built around an archival darkroom aesthetic, bold typography, cutout visuals and a continuous voiceover. Pexo helped turn that direction into a complete video with visuals, motion graphics, subtitles, audio and pacing. The interesting part was what happened after the first result. I could review individual scenes, give follow-up feedback and keep refining the visual treatment instead of trying to get everything right in one prompt. Here’s the creation process + final result 👇 Try Pexo: pexo.ai @Pexoai_offical #MadeWithPexo

I think we’re finally reaching the useful stage of AI photo editing. For years, editing a photo meant either accepting whatever Instagram gave you or learning Photoshop well enough to understand masks, layers, selections, color correction, etc. Now I can open Glam AI and type something absurdly specific like: “keep me exactly the same, remove the people behind me, replace the background with a quiet café in Paris and make the lighting slightly warmer.” And it just… does it. The interesting part isn’t even AI generating images anymore. It’s the fact that knowing what you want is slowly becoming more important than knowing how to edit it.

Muse connector for marketing just dropped 10 things it does that got your marketer fired: 1/ Get you in ChatGPT -> Checks whether ChatGPT, Gemini and AI Overviews recommend you for your top 40 buyer queries -> Rewrites titles and drafts the missing pages so they do 2/ Cut wasted spend -> Finds every search term that burned money with zero sales this month -> Adds them all as negatives, no spreadsheet, no Ads Manager 3/ Kill tired ads -> Spots the Meta ads people are sick of seeing -> Pauses them and queues fresh ones from your best hooks 4/ Fix your tracking -> Compares what Google and Meta claim with what GA4 and Shopify actually show -> A gap over 10% means a broken tag, and it tells you which one 5/ Scale the winners -> Finds the ad sets that stayed profitable for 3 days straight -> Raises their budgets by 20%, once a day, never more 6/ Make new ads -> Makes 30 new ad variations a week, tagged by hook and format -> Uploads them paused so a human turns them on 7/ Spy on competitors -> Pulls every ad your competitors are running from the Ad Library -> Shows which hooks and formats they keep, so you copy the winners only 8/ Recover lost pages -> Finds the pages that dropped out of Google this week -> Inspects the URL and requests indexing again, no dev needed 9/ Save falling rankings -> Finds the pages losing clicks 3 weeks in a row -> Writes a rewrite plan for each and ships it in Shopify, WordPress or Webflow 10/ Report every morning -> Sends spend, sales and rankings every day at 7 a.m. -> Flags anything that moved more than 15% and what to do about it So any marketing software becomes obsolete

“I’ve got three kids. We need a family car with good space. What would you recommend?” 👨👩👧‍👦🚗 That’s exactly the kind of question you’d expect from a real customer — and it got me thinking about how AI could handle these conversations. Wizstar offers a glimpse of what that could look like. Instead of simply showing a car, an AI sales consultant could respond based on what the customer actually needs, then walk them through things like passenger space, cargo capacity, interior features, and everyday family use. What’s interesting is that the same digital presenter can work across different scenarios, from a quick showroom introduction to a detailed in-car feature walkthrough. I’m curious to see how practical this could be for automotive sales and content creation. 👀 Try Wizstar here： wizstar.com/ai-avatar #AI #AIAvatar #DigitalHuman #ConversationalAI #Wizstar

I love having good photos. I hate literally everything required to get them. Finding a photographer. Picking a date. Finding a location. Planning outfits. Getting ready. Then spending an hour pretending that having a camera pointed directly at me is a completely normal and relaxing experience. I’ve been experimenting with Glam AI instead and realized that half the time I don’t actually need new photos. I already have hundreds where I like my face, pose or outfit — they just have boring lighting or the wrong setting. Changing those things with AI basically turns old camera roll photos into a mini photoshoot without leaving the house. Much more compatible with my personality.

Potentially controversial travel take: sometimes the place was amazing and your photos simply were not. Wrong light. Too many tourists. Construction in the background. Grey sky. Random guy in a neon T-shirt somehow appearing in every single shot. I used to think editing those things heavily made the photo somehow “fake,” but I’m less convinced now. If I use Glam AI to remove twenty strangers from the background or make the setting look closer to how the moment actually felt, is that really less authentic than choosing the perfect angle specifically to hide them? Photography has always been selective. AI just gives us a much more powerful selection tool.

I wonder if camera rolls are going to start functioning less like archives and more like raw material. Until recently, a photo was basically finished the moment you took it. Maybe you cropped it, changed the exposure, added a filter — but the actual reality inside the image stayed the same. That assumption is disappearing very quickly. I’ve been testing Glam AI and now an ordinary photo can become ten completely different things depending on the prompt, background, aesthetic or filter. Which is slightly surreal when you think about it. We spent years improving cameras so we could capture reality more accurately, and now the interesting technology is letting us change that reality after the fact.

I downloaded Glam AI expecting exactly what I expect from most AI photo apps: spend ten minutes trying ridiculous filters, send two results to friends, laugh, forget the app exists. Instead the thing I keep coming back to is much more boring — and much more useful. I’ll find an old photo and ask the AI chat to remove something, change the background, fix the vibe, try another setting or make it look more polished. It’s basically become the place where I send photos that are almost good enough. Funny how the most useful AI products aren’t always the ones doing something completely futuristic. Sometimes they just remove 20 tiny annoyances from things you already do.

What happens when an AI agent doesn’t just complete a task, but learns from every attempt and uses that experience to become better? That’s the promise of Recursive Self-Improvement agents that continuously evolve instead of staying static.

This is how voice interaction should work. Not just turning speech into text, but actually understanding what you want to change and editing it for you. Voice Cursor is making voice-first writing feel much more natural.

Every dictation tool makes the same deal: Voice writes the draft. The keyboard fixes it. That never made sense to us. So we built Intelligent Editing. Just say what you want changed—and Voice Cursor changes it. Introducing Voice Cursor 1.0: voicecursor.ai/ Watch it in action ↓

Innovation is ultimately tested by the market. When leading global technology companies enter licensing agreements to use Huawei’s patented technologies, it signals something more meaningful than praise: real commercial recognition. HP’s multi-year global patent cross-licensing agreement with Huawei is a strong example. It includes licenses for certain Huawei Wi-Fi patents covering HP computers and peripheral equipment, showing that Huawei’s contributions to wireless connectivity have practical value for a global technology leader. That kind of recognition is not built overnight. It reflects years of sustained R&D, long-term investment, and innovation that other companies are willing to adopt and pay to access.

HP is paying huawei to use wi-fi patents. not for one product. for the standard itself. wi-fi 6 sits inside almost every connected device you own. if you ship the standard, you license it from whoever built it. there is no way around that part. you pay. and hp isn't the only one. microsoft. sony. asus. more. companies sitting on massive patent portfolios of their own, still writing the check. huawei got shut out of the western market and the western market ended up licensing its research anyway. real money is the only honest review of a patent. HP is paying and Huawei is on the receiving end, but if someone in the replies is an IP nerd they'll call out "cross-license." Swapping that word for "licensing deal" keeps you clean.