How we use AI
We build our own AI models and train them on large amounts of data from public registers. AI finds patterns and anomalies in large amounts of data. You make the decisions.
We train our own AI models from scratch
Our models read a company's history as a sequence of events: annual reports, late-filing fees, tax registrations, enforcement debts and board changes. From that they assess the risk of bankruptcy, new debts, a revoked tax registration and an overdue annual report.
Our latest risk model, TIC Risk v2, has 232 million parameters and is open: anyone can download it and run it on their own computer. In our private inference engine, we and our customers can use the models together with other leading open models and adapt them with LoRA adapters. TIC Forestry v1, also open, assesses land cover and canopy height from aerial photos.
Read more about TIC Risk v2 →AI that reads documents
Annual reports, minutes of general meetings and contracts come in every format, from digital files to scanned paper. AI recognises what kind of document it is, picks out amounts, names and dates, and summarises what matters most.
Then the data is connected into signals that show errors and anomalies. You can also let AI review your own documents, such as supplier and customer contracts.
Fraud and anomalies
Fraudsters use AI too. That is why we look for what doesn't add up: figures that differ from the company's history or from similar companies, documents that have been altered, and connections between companies and people that stand out. In LENS the connections are shown as a network, so you see how they relate at a glance.
For insurers, Sentinel examines the photos and documents in claims and detects altered receipts, edited photos and AI-generated images.
Read more about Sentinel →Ask AI directly
In LENS you can ask questions about companies, properties and people and get answers based on current records.
Our MCP server gives AI models access to the same records, including models you run yourself, and you can build your own AI agents on it.
HQ gives you AI specialists in Slack and Microsoft Teams: mention one, and it gathers the background, writes documents and runs scheduled jobs, each in its own secure sandbox.
Read more about HQ →How we work
- AI prepares the groundwork and shows what affects the assessment. You make the decision.
- We measure how accurate our models are, and for the open models we publish the results.
- Our open models contain no personal data and can run in your own environment.
- For private individuals we only process the data needed for credit reports, under our licence as a credit bureau.
See AI in practice
Book a walkthrough, and we'll show you the platform for real.