Izlūks is a Latvian business intelligence platform. It pulls Latvian company data into one place, layers scoring and benchmarks on top, and makes the result queryable in plain language.
What Izlūks is
Izlūks brings together company registrations, board and shareholder history, financial indicators, beneficial owners, and sanctions screening. The data is wired so it can be walked through — jump from a company profile to its owner, from that owner to their other companies, from a company to its sector peers — instead of looking up one record at a time.
The interpretive layer
Above the data is the layer that interprets it. Risk scoring built from structural changes, missing filings, and links to sanctioned persons. Insolvency prediction. Credit-limit calculation. Peer comparison drawn from actual annual reports. AML reports compliant with NILLTPFN1 requirements, in the standard thirteen sections, ready to attach to a compliance file. Network analysis — link-finding between persons and companies, influence measurement, community detection across the ownership network.
This is the layer that answers what these numbers mean, not what they are.
Izlūks AI
Above everything sits the query layer: Izlūks AI. It is a surface where the user writes a question in natural language, and Izlūks returns an answer drawn directly from what it sees — companies, annual reports, sanctions, ownership links. The question's language is auto-detected — Latvian, English, or Russian — and the answer comes back in the same language.
Two modes run side by side. Quick Query (1 credit) — a single validated database query, returning rows and a short explanation. Best for fast lookups: top-N, filters, counts. Agent mode (3 credits) — multi-step planning, in which Izlūks AI chains tools in sequence: searches for a company, opens its profile, screens for sanctions, scores risk, draws a chart, cites sources. Sessions are persisted, so the next question keeps context. Failed runs and clarification turns refund credits automatically.
Three examples in Agent mode.
A compliance check. "Find Latvian SIA2 whose beneficial owner appears on the OFAC SDN list, sorted by 2024 revenue." The answer is a list of regkodi3, beneficial-owner names, and the underlying sanctions citation — ready to export or roll into a compliance file.
A journalist's query. "Which SIA liquidated in 2024 had revenue above EUR 5M, and what was their last reported profit?" The answer is a small list with links to each company profile, where board history, owners, and excerpts from filed annual reports are already in place.
A manager's question. "How does my company compare against five direct peers on profitability and debt-to-equity over the last three years?" The answer is a table and a chart, both built from the actual filed reports rather than from computed sector averages.
Three questions that, a year ago, were three working sessions: pulling extracts, stitching them together in Excel, manually checking against a sanctions list. Izlūks AI does them on the spot, because it sees the whole register at once — companies, persons, reports, sanctions, and the connections between them.
Every answer comes with direct citations to the underlying register entries and to the SQL queries used. When the data isn't there, Izlūks AI says so — no assumptions, no fabricated numbers.
A separate, smaller surface — Docs Chat — sits on the API documentation page. It helps developers find the right endpoints, parameters, and error codes, and it produces request examples in curl, Python, Node, or Go. The documentation itself is built so that AI coding agents can read it as well as a human can — Cursor, Claude Code, and similar tools can build an integration without extra hand-holding.
How to access it
The platform exposes two surfaces. A web app — search, company profiles, person profiles, risk reports, watchlists, PDF exports, Izlūks AI. A REST API — self-service keys with the ur_live_ prefix, an OpenAPI-documented endpoint set, sent on the X-API-Key header.
Pricing tiers:
- Observer — €0. Web app with a 3 AI-credit signup bonus. No API access.
- Agent — €15/mo. 50 AI credits per month, API access (50 req/h).
- Analyst — €29/mo. 200 AI credits per month, API access (400 req/h).
- Command — custom contract. Custom credit grant, integrations, SLA.
AI queries cost 1 credit (Quick Query) or 3 credits (Agent mode) — on every tier, including Command. Top-up packs (10 / 30 / 50 credits) are available alongside any tier.
The remaining issues use this platform; they don't describe it.
Notes
Footnotes
-
NILLTPFN — Noziedzīgi iegūtu līdzekļu legalizācijas un terorisma un proliferācijas finansēšanas novēršanas likums, the Latvian AML/CFT statute. ↩
-
SIA — sabiedrība ar ierobežotu atbildību, the Latvian limited-liability company form. ↩
-
Regkods — the Latvian commercial register's unique company identifier. ↩