The intelligent layer of the RiskantNO ecosystem.
Sigmata reads documents, checks MTPL, motor hull and property pricing, collects photos from the field, links physical objects to data and answers questions from your documentation. It is built into RiskantNO, GarantNO, RiskantNO RE and RestCAR — and also available standalone, via API.
Your data stays yours
AI models run on our own infrastructure (vLLM). Client documents are never sent to public AI services.
Every finding has a source
No black box: you can see which document, which processing stage and which rule version led to a decision.
People decide
AI prepares and checks; unclear cases go to manual review — with every finding in one place.
SafeLens
Sees what the human eye misses.
Send a policy, vehicle registration, accident report or any supporting document — SafeLens reads it, recognises what it is, extracts the data and compares it with the policy.
- OCR for Latin and Cyrillic — even scanned, rotated and blurry images
- Automatic document classification and field extraction
- Multiple documents and multi-page PDFs in one job
- Status of every document and stage, with reprocessing
PREMIUM_CLASSPremium class on the policy does not match the claims historyVIN_MISMATCHVIN on the registration differs from the VIN on the policyPostTarif AI
Every policy, checked.
After issuing, PostTarif AI checks whether a policy was priced correctly — for motor liability, motor hull and property. Every policy is compared against the tariff, regulations and real data, and deviations show up immediately, not at the audit.
- MTPL: premium class, bonus/malus, loadings and discounts under RS and FBiH rules, ID ⇄ VIN
- Motor hull: vehicle value, cover packages, deductibles, discounts and loadings
- Property: sums insured, rates, risk classes, location and surcharges
- Versioned rule sets — every decision knows which version it used
- No finding is lost: the check never stops at the first error
CORRECTLY_TARIFFEDREQUIRES_CORRECTIONMANUAL_REVIEWSQR
Every object has its own digital history.
A vehicle, device, building or document gets a signed QR code. Scanning it opens its card — with the records, files and actions you defined.
- Versioned records following a schema you design
- Files, subscribers and change notifications
- Actions on scan: report a claim, book a service, call your API
- Cryptographically signed tokens — the QR cannot be forged
SigmataScan
Documents in minutes, not days.
Your system requests documents via API; the client receives a link or QR code and photographs them with their phone — no app, no email, no scanner.
- Templates: define upfront which documents and sides you need
- Image processing and compression on the device
- Status, results and events via the partner API
- Retention periods and rule-based data deletion
Claim documents
Take photos of the requested documents. It takes about 2 minutes.
SigAssistent
Answers from your terms — with a source.
A chatbot built on RAG technology: upload policy terms, regulations and guidelines, and SigAssistent helps your users and clients — answering only from those documents and always citing the source.
- RAG: answers only from your documents, with citations
- Help for users and clients — FAQ and conversation scenarios
- Scanned PDFs through OCR
- Per-client isolation, budgets and prompt-injection protection
Embedded or standalone — your choice.
Inside RiskantNO products
In RiskantNO, GarantNO and RE, Sigmata works in the background — no extra setup.
Via API
A documented REST API (Swagger), per-client API keys and webhook notifications.
On your infrastructure
For the strictest requirements, Sigmata and its AI models can run on-premise.
Show us your documents. We'll show you what AI sees.
Whether you're launching a new product, replacing a legacy system or just exploring — get in touch. We reply within one business day.