m-doughty/LLM-Data-Inference

LLM-Data-Inference is a nascent Raku library (3 days old, 0 stars) that implements standard LLM orchestration patterns. While it provides valuable glue code for the Raku ecosystem—which lacks the mature tooling of Python (Instructor, PydanticAI) or TypeScript (Braintrust, TypeChat)—it offers no novel technical moat. The features it provides (JSON extraction, retries, and templating) are now increasingly handled natively by frontier model providers (e.g., OpenAI Structured Outputs, Anthropic Tool Use). Its use of Roaring::Tags for model routing is an interesting implementation detail but doesn't elevate the project beyond a utility wrapper. From a competitive standpoint, the project's defensibility is minimal because it relies on a niche language with a small developer base and competes against multi-billion dollar platforms that are rapidly absorbing these middleware features into their core APIs. Its primary value is strictly for the existing Raku community rather than as a general-purpose AI infrastructure play.