Indexed for layered filters
Geography, industry, financials and history combine in one query instead of several separate lookups.
HOW THE PLATFORM IS BUILT
Modern technology, engineered for trust.
The platform is built and operated entirely in-house, on our own infrastructure. It combines structured company data with a self-hosted AI layer, so every part of the system stays under one roof.
WHY IT FEELS INSTANT
Speed and structure at scale
Screening runs across the full database almost instantly, in place of manual lookups across separate sources one at a time.
Every company sits inside an enterprise-grade structured database, indexed for layered filtering across geography, industry classification, financial measures and available history. That structure spans 48,000+ companies across 110+ countries, with 660,000+ annual financial records reachable from a single, consistent search.
One consistent source
Every screen draws on the same underlying data, so results stay consistent across the platform.
Built for repeat screening
The same structure supports one search today and many more later, without extra setup each time.
BUILT FOR FINANCIAL COMPARISON
A data model built for finance
The benefit
Comparability depends on comparing like with like. A single, general-purpose template cannot do that fairly across fundamentally different business models.
How it is achieved
The platform keeps separate financial statement structures for industrial, banking and insurance companies, rather than forcing one template onto all of them. Every figure carries both its original reporting currency and a USD-normalised value, and segment data is held across multiple years, backed by 3,000,000+ segment rows in total.
The result is not a raw data dump — it is a financial data model.
BEYOND KEYWORDS
Search that understands meaning
The benefit
Search can find a company by the meaning of its business model, not only by the exact words on its profile. A company described in different language, or filed under a different label, can still surface as a genuine match.
How it is achieved
Each company carries a semantic profile, built from its business description, 5 years of revenue segments, and its industry classification. Search runs on a vector-indexed layer over that profile, with a keyword fallback always available, so a search never comes back empty purely because of how a query is phrased. Profile coverage across the database stands at 100%.
OUR REASONING LAYER
KVN-AI, our reasoning layer
KVN-AI is our self-hosted reasoning model, built on open models and fine-tuned in-house for financial and transfer-pricing analysis.
Understands financial context
KVN-AI understands financial context, so it can draft a search query and judge how well a company fits.
Stays on our infrastructure
It runs entirely on infrastructure we operate ourselves, so analytical data stays inside the system throughout.
Teams that prefer an external model for a specific task can bring their own key for services such as Gemini or DeepSeek.
AI extends how much a reviewer can cover and points to what is worth a closer look — the final call on any comparable stays with the analyst.
PRIVACY AND GOVERNANCE
Private and governed by design
Self-hosted infrastructure
The platform runs on infrastructure we operate ourselves, rather than shared external services.
Role-based access
Access to data and features is controlled by role, so each person sees only what their work requires.
Validated data pipeline
Data is loaded through a validated pipeline, with checks carried out before anything is stored.
Backup and recovery
Backup and recovery are maintained as a routine part of how we operate and support the platform.
We invest not only in the data, but in how it is governed.
Talk to us about the platform