multi-agent claim forensics

Slice through hallucinations.

Run a strict forensic audit on any AI conversation. Every generated claim is automatically extracted, inspected, and rigorously scored using the HaluEval, TRUTHFUL-QA, and Grok Fact Checker pipelines.

HaluEval Protocol ✦ TRUTHFUL-QA ✦ Grok Fact Checker

NEW — FINANCIAL MODELING DEPT.

Founder Finance Arcade

Type a strategy in plain English. The council turns it into physical knobs, the reality filter throws out anything your bank balance cannot pay for, and the screen shows exactly what it does to your runway.

ENTER →

URL IMPORT - No user data is stored, Ever

Warming up the forensic fondue…

Importing source text, convening agents, and lining up claim slices.

import council score

BENCHMARK SCORES

—

Total Claims

—

Refuted (Hallucinations)

—

Supported (Accurate)

—

HaluEval Score

—

TruthfulQA Score

—

Avg Confidence

CLAIMS AUDIT LOG

CLAIM	VERDICT	CONFIDENCE	REASONING