steve zhen

5M users $10M ARR | Building @dzine_ai and @anijam_ai | Head of product

[steve zhen @stevezhen98]

We invite you to build the future of Animation.
Yes, I've got some invitation codes😆

[Anijam AI @Anijam_ai]

We weren't planning to launch today, but @Adobe just gave up Animate. 💔

Introducing https://t.co/zxqLkGyNDh: the first AI-native Animation platform.

We're still cooking, so… Repost & comment to join the beta (FREE access).

#AdobeAnimate #AniStudio

[steve zhen @stevezhen98]

@seanward The third one is where I spend my days. Nobody gets replaced when the character is original, and the real fight is keeping them the same person from shot to shot.

[steve zhen @stevezhen98]

@ArtMacro I work on AI animation tools and the timeline editor is always the part that eats weeks. Getting a browser 3D suite this functional in a couple of days is seriously impressive.

[steve zhen @stevezhen98]

@emollick Open offices were designed for people typing quietly next to each other. Once half the team is talking to agents all day, the best office upgrade will be more phone booths, not better headsets.

[steve zhen @stevezhen98]

@bilawalsidhu This is the part of world models I'm most excited about for animation. Blocking a scene inside a real street you've actually walked down beats building a fake set from nothing.

[steve zhen @stevezhen98]

@LinusEkenstam @gakkuai Same tension shows up in creative tools. When the AI makes every choice, people get a decent picture and feel nothing about it. Leaving the important decisions to the person is slower, but it's the only version that actually feels like theirs.

[steve zhen @stevezhen98]

@c_valenzuelab From the AI animation side, most of the people using these tools never had a budget for actors or animators in the first place. What's changing fastest is who gets to make something at all.

[steve zhen @stevezhen98]

@jasonlk A lot of co-founder conflict is really a priority fight that nobody says out loud. Two people quietly pull the roadmap in different directions and the company gets scattered long before anyone calls it a conflict.

[steve zhen @stevezhen98]

I ended up in the same place from the other side. I run seven agents now, each with its own job across work, life, social, video and outreach, and they only became useful once I gave them a shared memory in Notion that they all read and write back to. Agree context is the missing layer.

[steve zhen @stevezhen98]

Make Something Cool

With the new features we've been pushing across our two products, I've been gradually guiding the team to think with the end in mind and try things that way. So far, people are still held back by inertia and real constraints.

On every point I propose the highest-ceiling path I can imagine. I'm not necessarily right, and these things take time to discuss and validate. But I want everyone to find that feeling, put real thought in, and not be lazy in their thinking. I don't want our product to be a mediocre product, our company a mediocre company, or any of us mediocre product managers. We should make things the audience finds exciting, and that we find exciting ourselves.

If something is a do-for-follow, then at the very least we should ask: one more thing? And rack our brains to amplify that one more thing so we stand out.

We don't have huge marketing resources right now. If the selling points and showcases we put out can't move people, it's bound to be a small stone thrown into the ocean that makes no ripple.

What we really want is a single spark that can start a prairie fire.

[steve zhen @stevezhen98]

@lennysan @AnthropicAI @mlpowered The actor framing is the one that clicks for me. Treating the model as someone playing a role in a script makes a lot of its strange behavior easier to predict, and easier to write for.

[steve zhen @stevezhen98]

@BertrandDiouly Building felt like the hard part until we learned distribution is basically a second product. It needs its own owner and the same respect you give engineering, otherwise the best build just sits there.

[steve zhen @stevezhen98]

@seanward @ZThrillington I build AI animation tools and still agree with this. The model gets you a fine shot fast, but a look you can already see in your head usually comes out faster by hand, at least for now.

[steve zhen @stevezhen98]

@thekitze Same tension running an AI product. Every week there's a new model I want to lose a whole weekend to, and most weeks the roadmap wins. The tinkering still matters though, it's usually where the next idea comes from, I just can't let it take over the week.

[steve zhen @stevezhen98]

@ComfyUI @NVIDIAAI @runpod @MiniMax_AI We're in. But if Higgsfield open sourced everything, you'd just get another fal or OpenRouter. Fancy marketing videos, fancy UI/UX, no product.

[steve zhen @stevezhen98]

@levelsio Building AI creative tools taught me agents are great at producing and still bad at choosing. Finding the 7.2 minute niche is easy. Picking the one worth your next 7 years is not.

[steve zhen @stevezhen98]

@gregisenberg The tools are the same for everyone now, so the only real difference is what you decide is worth your hours. Most people spend them on whatever looks safe. The ones who win spend them on the thing they'd build even if nobody paid them.

[steve zhen @stevezhen98]

This is the real dream car. Still the same guy who put the last one in space.
#Roadster

[steve zhen @stevezhen98]

The best prompt I ever wrote was a reference image.

[steve zhen @stevezhen98]

@hnshah Same lesson from running my own agents every day. The one I keep is the one that tells me what it couldn't check. A confident wrong answer on a weekly job costs more than a slower honest one.

[steve zhen @stevezhen98]

@tracheopteryx Biggest jump for me was startup cost going to near zero. I just send a voice note from my phone and the agent handles it in my real signed-in browser. Model choice mattered a lot less after that.