Marcus 3ΞY

Going on LinkedIn integration next. Should be enough for an MVP? What do you think?? #ai #assistant

@minchoi Who said this was AI?

This is literally my media stack now ChatGPT Images 2.0 → visuals Seedance 2.0 → animation Suno 5.5 → music 100% AI

shipped a fix at 2am last night. token limit kept crashing the agent mid-conversation. the bug was one line. finding it took 6 hours. that's the part nobody posts about.

i've been jumping between 4 apps for months. compoz, anna, trypure, a claude code package manager. everyone says focus on one thing. but building across products is how i learn what actually connects them. the patterns only show up when you're deep in multiple problems at once.

chatting with an agent is getting less like a chatbot and more like talking to someone who actually gets it. built this in compoz. the line between tool and teammate is getting blurry fast.

Hey @fal is the image to video Multi-Prompt been removed or am I on the wrong model ?

@Jason I’m in

We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added

long-term memory in compoz is making me lose my mind. you'd think "remember what the user said 5 minutes ago" would be easy. it's the hardest feature i've built. context windows, token limits, what to keep, what to forget. and if you get it wrong the agent sounds drunk.

@airjellyAI Most of apps trying to keep us on the screen while it should be the other way around. Life if not about screen. It's about enjoying time away from it.

🚨The world just started talking about AI Screen Context. We've been building something deeper — AirJelly can see your screen and remembers not just what you did, but who you are. No intent missed. No task overdue. 🪼Introducing AirJelly today, the world's first context-aware proactive agent, that lives on your desktop, grows with your context, and proactively get works done.

spent all morning debugging agentic flow in compoz. the conversation memory was leaking context between sessions. users were getting responses based on someone else's thread. found it. one misplaced cache key. 3 hours for one line of code. building AI tools is humbling.

@alxui_ux Yes I think most of people don’t get it. The animation orchestration is a creative proceess. U can’t blame them

90% Of Websites Have Bad Animation Design

I want two shirts, please. OpenAI just reset the bar for image generation. We thought Nano Banana was already pushing the limits. Turns out, we were still underestimating what “understanding” an image actually means. This new Image-Gen-2 model doesn’t just generate visuals. It understands space. Perspective holds. Layouts make sense. Objects relate to each other correctly. Interiors feel architected, not hallucinated. Every test I’ve been running got crushed. This is the first time an image model feels like it’s reasoning about a scene instead of just composing pixels. If you’re building anything visual — this changes the game. same basic prompt: help me turn this space into a modern interior Can u guess which one is from nano banana ?

@rom1trs Arcads

I built a GPT 2 workflow that turns one image into 20 viral ads GPT image 2 is the most powerful image model out there And you can use it to scale your ads campaign with Arcads workflow Comment Arcads to get the workflow now

Exciting news - GPT-Image-2 by @OpenAI has claimed the #1 spot across all Image Arena leaderboards! A clean sweep with a record-breaking +242 point lead in Text-to-Image - the largest gap we’ve seen to date. - #1 Text-to-Image (1512), +242 over #2 (Nano-banana-2 with web-search aka gemini-3.1-flash-image) - #1 Single-Image Edit (1513), +125 over #2 (Nano-banana-pro aka gemini-3-pro-image) - #1 Multi-Image Edit (1464), +90 over #2 (Nano-banana-2) No model has dominated Image Arena with margins this wide. Huge congratulations to @OpenAI on this major breakthrough in image generation! More performance breakdowns by category in the thread below.

@starks_arq @Bitcoinbeach Personal Show soon

Personal brands are completely going to change. We will start to see 1 - person Hollywood blowing up soon. Seedance 2.0 changed everything. Thank you @Bitcoinbeach - we’re opening the cinematic universe soon.

GPT-Image-2 is whild. I went from this old version to a new one with same prompt.

Anthropic allows OpenClaw usage again. From @openclaw docs.

@terencebuilds Yeah the light signal in the dark tunnel ! First time I sell something online. Would’ve been chasing around like a blind chicken for many other weeks if this didn’t happen.

@romainbey This hit hard. I've been there - chasing new ideas while ignoring what's already working. That surprise customer complaint is actually gold - proves product-market fit exists, just needs execution. Focus beats perfection every time.

I’ve been building nonstop. Learning from indie hackers, consuming tons of advice, probably more than I can apply. Still struggling to implement consistently. Jumping between ideas, trying to find the one with the most potential. A few months ago I shipped an app. It had bugs. Real ones. Hard to fix. Bills to pay, so I moved on and kind of forgot about it. Today I opened my emails and saw a complaint. A customer saying the app didn’t work, asking for a refund. First reaction: surprise. I didn’t even realize I had users. Then I checked. I actually had paying customers. That moment hit different. Frustrating because the product is broken. But also a huge relief, because it proves something important. People are willing to pay. The idea was already good enough. What’s missing isn’t another idea. It’s focus and execution. We don’t need ten projects. We need one that actually works. One we push all the way. One where we talk to users, fix every detail, and make it truly valuable. That alone is already a win. 🥳 first time mqking money on internet.

@openclaw Finally we can have some serious run !!

OpenClaw 2026.4.20 🦞 🧠 Kimi K2.6 support + provider-aware /think 💬 BlueBubbles iMessage sends + tapbacks fixed ⏰ Cron state/delivery cleanup 🔐 Gateway pairing + plugin startup hardening Less haunted. More useful. github.com/openclaw/openc…

@feifei_qiu 🤣🤣🤣

Just finished the "Cancel Subscription" feature. Please help test. 🥹