polym-sdr-leads-qualification

Automate lead qualification for Trigify CSV exports by running an LLM pass to extract evidence and assign sub-scores, then applying deterministic Python routing to bucket leads into qualified, review, or disqualified outcomes.

LLM evidence scoring for sales leads works by executing a single LLM pass per lead to extract supporting evidence and assign five sub-scores, which are then re-validated by deterministic Python guardrails to enforce campaign-specific ICP and disqualification rules.

Yes, you can use custom campaign ICP rules to score bulk leads by creating custom campaign packs that define specific scoring rubrics, signals, and hard disqualifiers for deterministic routing.

Yes, you need LiteLLM-backed environment variables including LLM_API_KEY and LLM_MODEL, with optional LLM_PROVIDER and LLM_API_BASE, to run the LLM evidence extraction required for lead qualification.

The best way to route leads into qualified and disqualified buckets is combining LLM evidence extraction with deterministic Python guardrails that re-validate scores against campaign-specific thresholds and hard disqualifiers.

A limitation of using LLMs for lead qualification scoring is that raw LLM output is subjective, requiring deterministic Python re-validation and routing to enforce campaign ICP rules, thresholds, and hard disqualifiers consistently.