Not AI-optional. Not AI-you-can-turn-off. There is no model in this product, and there is not going to be one.
Plenty of tools now advertise an AI toggle. That is not the same as not having one, because your work still passes through a product built around a model. Bitgrain has no neural network in it at all.
| Capability | Bitgrain | AI-first design tools |
|---|---|---|
| Generative models | None | Core to the product |
| Same input, same output | Always, by construction | No, results vary per run |
| Your work as training data | Never, it does not leave your device | Depends on the terms |
| Undo and refine | Yes, every control is a parameter | Reroll the prompt |
| Works offline | Yes | No, inference needs a server |
| Cost per export | Zero, it runs on your machine | Credits or compute |
This is a real constraint, not just positioning. There are things a generative tool does that Bitgrain simply cannot, like inventing an image from a sentence. The tradeoff is that nothing here is a slot machine.
No. It is the one commitment the product is built around, and adding a model would break the guarantee that the same settings always give the same result.
A toggle hides a feature. Here there is no model, no inference server, and no image upload, so there is nothing to switch off.
Classical image processing in a web worker: error-diffusion and ordered dithering, CMYK halftone screens with per-channel rotation, Sobel edge detection, riso separation, and FFT-based glitch.