Truth verification for the AI era
Dispel the fog.
Reveal the truth.
Enter a suspicious link, screenshot, message, or claim. Dispel turns ambiguity into a readable evidence report with calibrated trust scoring, risk framing, and recommended action.
FOG
INVESTIGATION
CLARITY
MIST DISPLACEMENT
The cursor and scroll path cut through the cloud field so the interface feels uncovered, not simply displayed.
LIVE INVESTIGATION CONSOLE
The homepage itself can run a real check.
NO MOCK REPORTS
Submit evidence. Watch the fog clear. Land in a real analysis.
Screenshot uploads route into the image pipeline. Message, claim, and URL checks route into the text engine. The animation is cosmetic; the investigation is real.
SYSTEM WALKTHROUGH
A short tutorial is built into the page, not buried in docs.
Submit anything suspicious
Links, screenshots, scam messages, or claims all enter the same investigation model.
Read evidence, not theater
Dispel explains what it inspected, what was found, and where the evidence still stays weak.
Act on the verdict
Reports end with guidance: verify, pause, avoid payment, preserve evidence, or share the receipt.
INVESTIGATION MODEL
Fog. Trace. Correlate. Resolve.
Uncertain material enters the chamber
Suspicious inputs begin as incomplete, emotionally charged, or manipulated surfaces.
Signals are isolated and labeled
Metadata, provenance gaps, scam rhetoric, synthetic indicators, and contradictions are separated into evidence categories.
The evidence matrix is assembled
Risk vectors are cross-mapped so a polished-looking scam cannot hide behind a single strong signal.
The result becomes readable action
A final report condenses the investigation into trust classification, score, evidence, and next step.
EVIDENCE FIRST
Clarity comes from signal, not from louder branding.
DISPEL
READS SIGNALS
Provenance status
When origin data exists, it is surfaced. When it is absent, the report says so without pretending otherwise.
Manipulation indicators
Visual, linguistic, and media-level cues are normalized into a calm, defensible score model.
Scam patterns
Urgency, secrecy, impersonation, payment pressure, and coercive framing are translated into readable warnings.
Trust classification
Reports resolve into trusted, likely real, mixed, suspicious, or high risk, with unknown states preserved when needed.
SCENARIO THEATER
See how misleading content unravels once the fog is forced to move.
CASE 01
The recruiter who wants money before onboarding.
A polished hiring message asks for identity documents, promises immediate onboarding, and pressures the user to pay for equipment outside normal process.
- Authority impersonation pattern
- Urgency before verification
- Payment request before legitimate onboarding
- Source mismatch against public company workflow
CASE 02
The voice note built to bypass your caution.
An emotional emergency call demands immediate transfer, but identity proof and source authenticity cannot be established from the available evidence.
- Urgency and coercion spikes
- Identity remains unknown
- Origin chain absent
- Recommended action: verify through a separate contact path
CASE 03
The listing that looks credible until the evidence is compared.
Seller screenshots, product imagery, and message history appear clean at first glance, then collapse once off-platform payment pressure enters the story.
- Manipulation risk in the supporting image set
- Payment outside protected rails
- Contradiction across seller claims
- Low provenance confidence
CASE 04
The guaranteed upside ad with no trustworthy origin.
Celebrity-coded trust cues, synthetic promotional tone, and unsupported certainty are used to trigger action before due diligence can happen.
- Extraordinary claims without evidence
- AI-style promotional language patterns
- Origin remains unverifiable
- Recommended action: do not send funds from ad copy alone
PRODUCT SURFACES
Consumer clarity, creator protection, and operator-grade trust tooling.
Persistent investigations with saved history and full report pages.
Shareable verdict artifacts that explain risk in under five seconds.
Programmatic scoring for platforms, trust and safety, moderation, and workflows.
FAQ
Calm answers to the obvious questions.
Can Dispel prove something is fake?
No. It evaluates what can be defended from available signals and marks uncertainty directly when the evidence is weak.
Is this a lie detector?
No. Dispel focuses on provenance, manipulation indicators, scam risk, and claim consistency. It does not claim mind reading or biometric truth detection.
How does the homepage tool relate to the full app?
The homepage runs real submissions, then hands the user into the full analysis detail view so history, reports, and upgraded workflows stay available.
FINAL CTA
Trust the evidence, not the performance.
Run the investigation before you trust, share, or pay.