HuMetric
agentic metric engine, ai, llm, entity intelligence
Start new thread

HuMetric - agentic metric engine, ai, llm, entity intelligence

by•
HuMetric turns unstructured text reviews, support tickets, social mentions, CRM notes into calibrated, temporally-decaying entity metrics. A multi-agent LLM pipeline (extractor → curator → ranker) reads raw signals and produces structured, confidence-scored metrics per entity, with full audit trails back to source text. Domain-agnostic: define your own Metric Pack in YAML for any entity type no retraining, no fine-tuning. Open source.Selfhostable. Multi-tenant with row-level security by default.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt! 👋 I built HuMetric because every "customer intelligence" or "entity scoring" tool I looked at was either a black box or locked to one specific use case (support tickets, or reviews, or leads — never both). HuMetric is domain-agnostic: you describe what you want to measure about an entity — a customer, a supplier, a candidate, a hotel guest — in a simple YAML "Metric Pack," and a multi-agent LLM pipeline (extraction → curation → ranking) turns raw text signals into calibrated, confidence-scored metrics that decay over time as new signals arrive. Everything is traceable back to the source text, multi-tenant by default (Postgres row-level security), and fully open source — self-host it or point it at your own Anthropic/OpenAI/Google/DeepSeek key. Would love feedback, especially from anyone doing entity scoring or lead/customer intelligence today — what's missing, what would make you trust the output?