Here is the recipe.
Go build.
The architectural patterns behind every flagship on this platform — with the live demos that use them and the code to start from. These are reference patterns, not products: rough edges are documented, not hidden.
Multi-head decision layer
Many classifiers, one frozen backbone: task-specific heads share a single LFM2.5-350M forward pass.
Token-classification redaction gateway
BIO-tagged token heads that find and redact sensitive spans at the boundary — one pattern, many industry skins.
ColBERT late-interaction retrieval
Per-token MaxSim scoring instead of one pooled vector — retrieval that survives catalog drift.
Constrained decoding
Structural validity is the decoder's job: GBNF grammars and trie logits processors make invalid output impossible, not unlikely.
Encoder on a frozen backbone
A small modality encoder projects structured events into a frozen LFM2.5-350M; LoRA plus task heads read out decisions.
The LEAP fine-tune loop
Task → data → train → eval → deploy: a new specialist head in minutes, gated by evals before it serves.