Hey, very interesting product. I've been making motion graphics as well and absolutely prefer this pathway, deterministic, generally makes less mistakes and most importantly editable and debuggable. That said, I think since the Opus 5.5 drop, the ability of vanilla Claude setup creating motion graphics with remotion, or combining computer use with after effects, has taken agentic motion graphic a major leap and I haven't really faced anything major and been able to get most things done within one shot. Anyways, wondering what's your thoughts on this and where you see the long term durable advantage as well
I agree, the models have gotten really good, and Klistra uses a similar pipeline to Claude + Remotion under the hood. It doesn't use Remotion, but it's currently running on Opus 5.5.
What I tried to do with this project was make it more accessible and integrate all the supporting tooling into one studio with a purpose-built graphical interface.
If you're a developer who's already got your own Claude + Remotion workflow set up, you might not need Klistra. It's mainly for people who'd rather spend the time on the idea than on the setup. That said, I use Klistra myself even though I had the Claude workflow set up before, because the supporting features save me time, and with the Claude connection you get a bit of the best of both worlds.
I think the biggest downside to Klistra right now compared to the manual pipeline is the cost. If you already have a Claude subscription, it's going to be more expensive, since Klistra has to pay API prices. A first 15–20 second video usually takes 50–150 credits, roughly $2–9 depending on plan. The Claude integration was an attempt to partly fix this, but the builder agent itself still runs on the API.
My hope for the future is that models will get better and cheaper, so this won't really be an issue. I also have plans to gather my own data and try to fine-tune an open-source model, which could be both better and cheaper, but so far I've focused on getting the product out there.
Curious what you mostly make, and where your setup still falls short?
Hey, very interesting product. I've been making motion graphics as well and absolutely prefer this pathway, deterministic, generally makes less mistakes and most importantly editable and debuggable. That said, I think since the Opus 5.5 drop, the ability of vanilla Claude setup creating motion graphics with remotion, or combining computer use with after effects, has taken agentic motion graphic a major leap and I haven't really faced anything major and been able to get most things done within one shot. Anyways, wondering what's your thoughts on this and where you see the long term durable advantage as well
Thanks!
I agree, the models have gotten really good, and Klistra uses a similar pipeline to Claude + Remotion under the hood. It doesn't use Remotion, but it's currently running on Opus 5.5.
What I tried to do with this project was make it more accessible and integrate all the supporting tooling into one studio with a purpose-built graphical interface.
If you're a developer who's already got your own Claude + Remotion workflow set up, you might not need Klistra. It's mainly for people who'd rather spend the time on the idea than on the setup. That said, I use Klistra myself even though I had the Claude workflow set up before, because the supporting features save me time, and with the Claude connection you get a bit of the best of both worlds.
I think the biggest downside to Klistra right now compared to the manual pipeline is the cost. If you already have a Claude subscription, it's going to be more expensive, since Klistra has to pay API prices. A first 15–20 second video usually takes 50–150 credits, roughly $2–9 depending on plan. The Claude integration was an attempt to partly fix this, but the builder agent itself still runs on the API.
My hope for the future is that models will get better and cheaper, so this won't really be an issue. I also have plans to gather my own data and try to fine-tune an open-source model, which could be both better and cheaper, but so far I've focused on getting the product out there.
Curious what you mostly make, and where your setup still falls short?
[flagged]