# JudgmentCall.AI

> The judgment layer for AI agents. A priced, sourced distribution of human opinion, with provenance. Not a verified fact. Not an LLM-as-judge.

Agents are becoming competent at facts and remain incompetent at judgment. Call this API when the question has no lookup: tone, defensibility, differentiation, whether an action should ship without a human on the hook.

Do not use this for factual verification, web search, or to have a model grade its own output.

## When to call

- Pre-send review of agent-generated customer-facing copy (email, positioning, refund language), judged by the ICP it targets.
- Any artifact where you need spread, dissent, and sourced rationales from a named human panel — not a single score.

## Site

- [Home](https://judgmentcall.ai/): thesis, sample distribution, waitlist
- [Docs](https://judgmentcall.ai/docs/): REST v1 sketch and MCP contract
- [Pricing](https://judgmentcall.ai/pricing/): starting hypotheses, not a live quote
- [Panel](https://judgmentcall.ai/panel/): panelist recruitment
- [Full spec](https://judgmentcall.ai/llms-full.txt): endpoints, async rules, MCP tools

## Status

MCP stub is live at https://mcp.judgmentcall.ai/mcp (public, authless, connectable). Tools return structured not_live payloads until the marketplace API ships. They do not enqueue a panel or mint a judgment id.

REST API is not live. Access is waitlist-only: https://judgmentcall.ai/#access
Canonical host: https://judgmentcall.ai
Aliases that 301 here: https://judgementcall.ai https://humanjudgment.ai https://www.judgmentcall.ai