Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
Category: ai-ml
Tags: ai-agents, llm-tooling, model-optimization
Score: 7.5/10 (Innovation: 8, Technical: 6, Documentation: 9, Utility: 7)
Tier is an adaptive tool routing system for AI agents that dynamically adjusts tool descriptions and parameters based on the size/capability of the underlying LLM. It's interesting because it addresses a fundamental bottleneck in agent tool selection—navigation, not selection—by presenting tools differently to small vs. large models, achieving significant accuracy gains and token savings.
Target audience: ai engineers, ml researchers, backend devs building agentic systems
Repository: https://github.com/yantrikos/tier · Python · MIT
View on Hacker News