Fraud prevention as SaaS: the vetting platform that evolves with every new scam
Phantom carriers use forged documents to win loads and then disappear with the cargo. Cargo Check AI® brings the key steps of selecting subcontracted carriers together in a standardized digital process. As technology partner, we worked with the Cargo Check AI® team to build the SaaS platform, from technical concept through architecture, backend and frontend to go-live, and today we support them in an advisory role on security and quality assurance.
A platform we can develop further on our own
For us, the key was not simply to digitize our own vetting methodology, but to translate it into a platform that can evolve flexibly. Together with Digit-One, we created a modular technical foundation on which we can now develop Cargo Check AI® further on our own and integrate new vetting mechanisms.
Mag. Matthias Guderjahn
CEO & Co-Founder @ Cargo Check AI®
carrier checks
red risk assessments
weeks until the MVP
mins to the check result
When fraudsters look like real companies
Cargo crime has become professional. Criminals increasingly pose as legitimate carriers, with plausible company data, documents and contact details. Especially when loads are assigned at short notice, it is becoming harder and harder to tell real companies from carefully prepared fake carriers. The Cargo Theft Tactics and Trend Report 2026 by Munich Re and BSI describes exactly this shift: instead of force, criminals increasingly rely on the right paperwork, forged identities and inside knowledge.
A process that has to keep pace with the scams
This is where Cargo Check AI® comes in: the selection process becomes structured, standardized and traceable. Dispatchers upload a carrier’s relevant documents, such as the company register extract, EU license, CMR insurance confirmation, letterhead and email history, and the platform cross-checks them against each other and against its own reference data. The real challenge was the technology behind it: scams change constantly, so the vetting logic has to adapt quickly too. At the same time, the MVP had to go live within a few weeks while still being built on an architecture that scales from the pilot customer to thousands of users.
Concept first, then code
Before writing the first line of code, we defined the architecture in a compact concept phase and ran through the vetting process in spikes using sample documents. The Cargo Check AI® team created the design themselves, and we reviewed every iteration for feasibility and effort. The result was a backlog cleanly split into MVP and V1, plus an API definition that bound frontend and backend together from day one. The concept was completed at the end of October 2025, and the MVP sprint followed immediately.
Stable core, modular vetting logic
The central architecture decision: we split the system into a stable core and a modular vetting layer. The core consists of a Next.js frontend with Tailwind, a Laravel backend with database and file storage, and Clerk for login, sessions and user management. The actual vetting logic lives outside the core in independent check engines that connect to external AI services where needed. This keeps the core lean and stable while the checks evolve independently.
AI sorts what gets uploaded
Users simply drag their documents into an upload field. On upload, the backend classifies each file with AI and assigns it to the matching document type, and in the interface the files visibly move into place. The classification service is abstracted and can be swapped for a different model if needed.
Check engines: modules instead of a monolith
Each check engine is an independent module that works with the backend through a clearly defined interface. When a check starts, all active modules run in parallel and their results feed into one overall assessment. Which technology powers a module makes no difference to the system. New vetting modules and logic can be adapted and extended largely independently of the core system.
Every check stays traceable
When liability and insurance are at stake, every assessment has to be backed up. For users, there is just the check with its documents, results and traffic light. Check runs are versioned and documented in a traceable way within the applicable retention and deletion rules.
Security as the foundation
Security and data protection were part of the architecture from the start. Every company sees only its own checks and documents, and tenant separation is built directly into the interface. Sensitive data and documents are shielded from the outside, and access is limited to what is strictly necessary.
Run like a big product
Even an MVP deserves a clean workflow. Production and staging systems run in parallel, and releases go live through automated pipelines. New vetting modules are first set up in the test system and only activated for real users after approval.
The first checks
At launch, three check engines went live, plus a template for all future ones:
- Address and company name: cross-check across all uploaded documents.
- CMR insurance: review of validity, coverage details and plausibility against the relevant criteria.
- Letterhead and email: cross-check of company and contact details from the letterhead and email history.
All results come together in a risk traffic light: green means unremarkable within the checked criteria, yellow means manual clarification is needed, red means
The result
From concept completion to the MVP going live on December 15, 2025, it took around six weeks. Since the V1 launch on June 7, 2026, dispatchers have been vetting their subcontracted carriers on Cargo Check AI® in production. Instead of phone calls and spreadsheets, they get a result usually in under five minutes, with every single step documented. To date, more than 1,300 carrier checks have been run, and more than 400 of them ended with a red risk rating, meaning significant risk signals that should be clarified before assigning a load. And because core and vetting layer are cleanly separated, the platform grows without the architecture having to be rethought.
What’s next
Since the MVP went live, we have continued developing the platform together in an agile way and have already added several new features, including a dashboard with all checks and a downloadable PDF check report. In parallel, the Cargo Check AI® team has built additional check engines themselves based on our template, exactly as the architecture intends. The modular architecture makes it possible to add further vetting mechanisms, integrations and enterprise features step by step. This brings Cargo Check AI® closer to its goal: every load assignment checked in a structured way and fully traceable.
FAQ
As technology partner, Digit-One worked with the Cargo Check AI® team to build a modular SaaS platform for AI-powered carrier vetting, from technical concept through architecture, backend and frontend to go-live. The vetting logic and product development lie with Cargo Check AI®, and today we support the platform in an advisory role on security and quality assurance.
That mainly depends on how clear the scope is before development starts. With a compact concept phase that defines architecture, backlog and interfaces, a few weeks is realistic. For Cargo Check AI®, it took around six weeks from concept completion to the MVP going live.
By separating what has to stay stable from what changes constantly. Login, data storage and the interface form a lean core, while the domain logic lives in independent modules. At Cargo Check AI®, this means new vetting modules and logic can be adapted and extended largely independently of the core system.
Specifically where it takes work off people’s hands, and abstracted so the model stays interchangeable. At Cargo Check AI®, AI automatically assigns uploaded documents to the matching document type. The service behind it can be swapped for a different model if needed, without rebuilding the platform.
With tenant separation that is built into the architecture from the start rather than added later. At Cargo Check AI®, every company sees only its own checks and documents, and the separation is built directly into the interface.
By keeping every result linked to the data it is based on. At Cargo Check AI®, check runs are versioned and documented in a traceable way within the applicable retention and deletion rules. This matters most where liability and insurance are at stake.
Yes, if the architecture is designed for it and the team gets the right tools. At Cargo Check AI®, the team has built additional check engines themselves based on our template, alongside our joint development of the platform.