ScoRISKTrust Intelligence
Private previewearly access open

Anyone can score it.We show the work.

A trust intelligence engine that scores how far any claim can be relied on — and hands you every signal, every source, and every step behind the number.

Request early accessSee the proof chain
No black boxFull proof chainExportable evidence report
app.scorisk.io / analysis / a3f72c9b
Live

Analysis Overview

scoring v9 · vector v2.1.0
Detailed breakdown
Strongly Aligned
WHO declared COVID-19 a global pandemic in March 2020.
hash: a3f72c9b1e4d8f21… · verified · 12 verifications
91
TRUSTMARK
94
CONFIDENCE
Truth signalsD1Sub-component values · PSD §1.5
Veracity confidence90
Evidence strength86
Consensus alignment93
Source corroboration88
Contradiction score18
Source IntegrityD2
who.int
Tier 1 · transparency 0.96 · history 0.98
Highly Reputable
0.94
Built for teams who must justify a callBanks & insurersTrust & SafetyGovernmentThreat intelligenceAI platformsNewsrooms
The problem

You can’t defend a decision you can’t trace.

Generic AI chatbots hand you an answer with no traceable reasoning. When an editor, an auditor, or a regulator asks “why did you rate it that way?” — a black box has nothing to show.

Verdicts with no receipts

A score with no traceable reasoning is just an opinion. You can’t reproduce it, cite it, or stand behind it.

Sources treated as equal

Most tools flatten a Tier 1 authority and an anonymous blog into the same signal. Origin credibility gets lost.

Nothing to hand an auditor

When the decision is challenged, you need an evidence report — not a screenshot of a chatbot reply.

How it works

Every score is a proof chain — not a guess.

Four stages turn a piece of content into a number you can defend. Nothing is inferred in a step you cannot inspect, and the same input always produces the same explainable result.

01

Extract

The text is broken into atomic claims — each one a single assertion that can be checked on its own, rather than a paragraph that is partly right.

02

Ground

Each claim is searched against the open web and the evidence base. Independent sources are collected, tiered by reliability, and corroboration is separated from contradiction.

03

Score

The five TrustMark dimensions are scored and weighted into a single number, with the contribution of every dimension recorded rather than collapsed.

04

Persist

Score, evidence, and reasoning are written to a versioned trail. Re-run it next month and you can see exactly what changed, and why.

Proof ChainExample
Claim extraction
1 atomic claim identified · LLM
Web search grounding
12 results · Brave Search · 3 Tier 1 sources
Fact-check · LLM signals
Verified · confidence 0.94 · no contradictions
D1 scoring · Truth Alignment
Score 91 · weight 0.35 · PSD §1.5
TrustMark persisted
CRV updated · audit trail written

Not a black box

Every signal, every source, every step is surfaced — a complete chain from raw evidence to final score.

Defensible & reproducible

The same input always produces the same explainable result — to a colleague, an editor, or a regulator.

An auditable evidence report

Export the full trail, attach it to a decision, publish it, or hand it to an auditor. That's the hero deliverable.

Early access

ScoRISK is in private preview.

Request access and we’ll be in touch. Tell us how you work today — we’re onboarding teams who need defensible credibility evidence.

No spam, ever — we review every request
All five TrustMark dimensions, live
Early access shapes what we build next
Flexible credit-based pricing — details at early access.
No launch dates promised. We review every request individually.