Skip to content
AI scam-investigation agent · grounded in MongoDB Atlas

Spot ticket-resale scams before you pay.

Paste a listing, seller DM, screenshot, PDF or link. A multi-step AI agent investigates it — grounded in MongoDB Atlas and the MCP server — returns an evidence-backed risk verdict, and hands you an exportable evidence packet plus the next actions to take. Built for 2026 World Cup resale fraud.

How it works
94%of seeded scams caught in evaluation
MongoDB Atlas Vector SearchGemini 2.5MongoDB MCPChange Streams

Decision-support only, not a guarantee — verify independently. · Synthetic demo data.

How it works

A five-step agent investigation

Ingest

listing / DM

Agents

Gemini 2.5

MongoDB

Atlas + MCP

Verdict

evidence-backed

01GEMINI

Extract the signals

Gemini pulls the price, seller handle, any domain, the payment method and urgency cues out of the raw listing or DM.

02VECTOR

Hybrid search the corpus

A vector pipeline and a full-text pipeline run over a corpus of known scam patterns, blending semantic and keyword relevance.

03MONGODB

Score & verify the verdict

An aggregation scores the risk, a rule checks for official digital transfer, and Gemini writes an evidence-backed verdict.

For portals

Drop-in protection for any resale marketplace

Embed the TicketGuard risk check on a listing page with a single tag, or call it programmatically — the same investigation is exposed as an MCP tool so agents can verify a listing inline.

One-line embed

<script src="https://cdn.ticketguard.dev/widget.js"></script>

Renders a risk badge next to any listing and opens the full investigation on click.

Exposed as an MCP tool

tool: check_listing(text) → verdict

Any MCP-capable agent can call check_listing to score a listing and get the same evidence-backed verdict inline.

About the data

Synthetic demo data only

Every example, seller handle, domain and risk verdict in this demo is synthetic and generated for illustration. TicketGuard surfaces risk signals consistent with documented patterns — it is decision-support, not a guarantee. Always verify a seller and pay only through official, protected channels.

94%

seeded-scam recall

2,480

corpus listings

14

scam patterns tracked

3.9s

median investigation

TicketGuard · MongoDB Atlas Vector Search · Gemini 2.5 · MongoDB MCP · Change Streams

One engine · every channel

Bring us the offer — from anywhere

However a suspicious ticket reaches you, hand it over and the same Gemini + MongoDB Atlas agent investigates it — and returns an evidence-backed risk verdict in seconds.

Screenshot a DM

Live

Got an offer in an Instagram, WhatsApp or LinkedIn DM? Screenshot it and drop it in — Gemini reads the seller handle, price, payment ask and urgency straight off the image.

Paste the message

Live

Paste the raw seller message or listing text. The agent extracts every entity and runs the full multi-step investigation.

Upload ticket evidence

Live

Drop in a ticket PDF or photo. We read the barcode/QR, inspect the file's metadata for tampering, and flag duplicates — risk signals, not an authenticity guarantee.

Paste a link

Live

Drop a resale URL. We fetch the real page, keep the host that actually served it, and score the listing — typosquatted domains get flagged.

We never read your inbox or DMs

By design. Auto-scanning a private Instagram/LinkedIn inbox is against their terms, and silent email access needs a heavy security audit. So you stay in control — you hand us the one message you're unsure about, and nothing else. Forward-to-email and native phone Share are on the roadmap.