Modern AI systems are overwhelmingly delivered as opaque, all-or-nothing monoliths. When something fails—or costs too much, or drifts—engineers are forced to guess from the outside.
The Case for Composable Primitives
Subsumptive is built on a different premise: intelligence should be assembled from discrete, inspectable building blocks that developers can understand, benchmark, and replace independently.
- Tokenizer Block: Decouples vocabulary and segmentation from model weights.
- Memory Block: Manages structured context and episodic state as an explicit layer.
- Model Adapter Block: Provides hot-swappable LoRA and routing wrappers.
- Runtime: Coordinates execution graphs with deterministic fallback paths.
Adding Images to Posts
Place any image in public/blog/ and reference it either in the frontmatter (coverImage: "/blog/your-image.png") or inline in Markdown:
