Shanny Keidar Goldschmid

Intermittent variable rewards are one of the core drivers of engagement across entertainment. Short dramas use cliffhangers. Sarah J. Maas and Rebecca Yarros use emotional tension, delayed reveals, and unpredictable relationship payoffs. The audience never knows exactly when the payoff will happen: the confession the betrayal the plot twist the romantic payoff That uncertainty creates anticipation. And anticipation sustains engagement. The lesson for games: Strong retention loops are not only about progression or rewards frequency. They’re about creating emotionally meaningful uncertainty around outcomes players care about.

An Artist | Short Movie It’s been a while since I wanted to try SD2 with an animation style. So I wrote a little story and made it along the way. Made with @runwayml & @dreamina_ai

Beginner video: How to install & use Grok Build (made for non-technical SuperGrok and X Premium+ users) I got so many questions from friends, so I made this simple step-by-step guide. You’ll see exactly how to: • Install Grok Build in seconds with one command • Create real websites • Use Grok Imagine to auto-generate images & videos • Run multiple projects at once in different folders Grok even runs commands for you. No coding experience needed. Watch the full walkthrough 👇

Prototyping game assets directly with Grok @imagine

Can't wait to accidentally sacrifice my Queen for absolutely nothing🥸

AI systems expand relentlessly. Design is the art of selection. As products become AI-native, UI/UX designs can no longer be just fixed artifacts we ship. They should become living systems that continuously evolve - shaped by behavior, explicit preferences, context, memory, and personal history. Hyper-personalization can drive strong habits and loyalty. But push too far and you trigger reactance (users feel manipulated, like their privacy has been violated, and they disengage). Essentially, UX shifts from composing static screens to architecting adaptive narratives. The future designer is a systems creator: defining the logic, guardrails, memory layers, and behavioral consistency that keep these evolving products coherent, trustworthy, and human. Practical moves: • Bake in transparency and easy overrides so users stay in control • Set explicit boundaries to prevent creepy over-personalization • Test relentlessly for healthy engagement, not addiction or bias creep • Treat the AI as a capable, humble collaborator - always augmenting the user’s agency and judgment, never replacing or overriding it. We don’t just ship static designs anymore. We author dynamic systems with memory and story.

the woods are not safe small tip for you guys: now, I write my prompts in chinese on Seedance 2. Believe it or not, it’s more accurate and has fewer restrictions I let my man @ekello make the soundtrack for this one

2.5 million people use @ComfyUI monthly. One developer just added 150 professional tools to their toolkit for free: animal animation, VFX compositing, mask diagnostics, the kind of capabilities studios build internally and never release.

How people adopt and interact with AI is still one of the most overlooked drivers of where technology actually goes next. A person’s sense of agency (their perceived ability to control and shape outcomes) strongly influences how they engage with AI. Professionals might feel strong agency in their areas of expertise and view AI as a powerful amplifier of their own capabilities, yet that confidence tends to diminish when they move into unfamiliar domains. They’re more inclined to experiment, customize, and build solutions using open-source models, no-code platforms, and flexible tools. By contrast, those with lower perceived agency in a given area seem to gravitate toward polished, ready-made services where the complexity is abstracted away. They place higher value on convenience, reliability, and expert-managed solutions - often showing greater willingness to pay for that simplicity. The same pattern seems to emerge at the organizational level. Companies with a strong internal sense of agency are more likely to invest in building proprietary AI systems and maintaining data control, while others opt for established SaaS platforms and managed services to minimize risk and operational burden. >> What I find particularly compelling is the opportunity this creates for brands and product teams. Companies that thoughtfully address both mindsets (segments) - offering flexible “build-your-own” pathways for empowered users alongside intuitive “set-it-and-forget-it” solutions - may see stronger adoption, reduced churn, and deeper customer loyalty. >> Agency isn't fixed. It's context-dependent and can be grown. Good onboarding, education, and progressive disclosure (start simple, unlock advanced controls) can move people from low-agency to higher-agency customization. * These are my current reflections based on patterns I’ve been noticing across recent consumer & enterprise contexts.

@benln Tel Aviv 🇮🇱

Cafe Cursor, round two. Where should we go first? • Paris • London • New York • Bangalore • Madrid • San Francisco • Berlin • São Paulo • Bogotá

Its hard to hide from these truths but MSM and the Left do their best to obscure and ignore.

This AI workflow is wild 🔥 I used ChatGPT Image 2.0 to create a perfect animated scene. Then Seedance 2.0 turned it into a full cinematic animated video in minutes Super simple workflow below 👇

As promised, prompts below ✨

What if each playthrough felt uniquely made for you?

@keepitHumanUK ‘Generation’ here refers to API wrappers around generative models. If you’re an artist using AI, there’s no real need to pay a middleman. you can run models locally and build your own workflows for free (:

@KGshanny Agree and most artists have been testing how to implement AI tools in their creative workflows for the last 10 years. I was reacting to the way you frame it ‘Stop Paying for 2D Asset Generation’ 😥

Stop Paying for 2D Asset Generation. Create Game-Ready Sprite Sheets Locally with ComfyUI👇 * Workflow A (any image model online → Wan 2.2) * 1. Generate a character (Gemini, ChatGPT, Grok, etc.) 2. Feed into Wan 2.2 (img→video) 3. Add Video Slice → RMBG (background removal) → Clean animated frames, engine-ready * Workflow B (Flux + style LoRA → Wan 2.2) * 1. Create a consistent full-body character in Flux 2. Apply a game style LoRA (to lock art direction early) 3. Pass to Wan 2.2 (high + low noise UNETs + 4-step LoRAs) 4. Slice + remove BG → Strong style consistency outputs What actually matters: > Prompt motion explicitly: walk cycle, opposing limbs, stable torso, no camera > Lock style early (Flux + style LoRA) before animation > Keep it structured: image → video → frames → transparent sprites From here, it's all about benchmarking seeds and tweaking weights. These might take more effort than using a paid wrapper, but building your own workflows and using models directly (local or API) gives you true creative freedom and direction control. #gamedev #comfyui

In my view, AI acts as an amplifier of creativity. With a deeper understanding of the technology and its evolving capabilities, artists can express their unique vision more effectively, accelerating both the pace of creation and the ability to produce at scale. Importantly, running local models enables this without ongoing costs. giving creators the ability to work freely, avoid paying for model wrappers, and retain full control over their workflows, tools, and output🙆‍♀️

@KGshanny Well there are lots of great 2D and 3D artists out there who deserve better.🤗

This breaks the mold for AI-animated storytelling 👏👏👏

I built my own custom tool for generating ad creatives. Creating visual concepts and testing ad variations used to take forever... Now I just throw in raw game assets & ideas🤓 #gamedev

Customization + creative expression = emotional stickiness → better retention Custom Codex pets aren’t just cute. They’re shareable memes waiting to explode (virality flywheel activated). OpenAI is playing the long game: • Turn a powerful coding agent into something you actually feel attached to (daily habit unlocked) • Spark massive user-generated content and sharing loops This isn’t fluff. It’s smart UX + retention + growth engineering in one adorable package🐾