Automating company due diligence (KYB) in Lithuania with an AI agent

You can automate most of a Lithuanian company due-diligence check with an AI agent connected to ImonesMCP. In one conversation the agent can confirm a company exists and is active, check whether it is VAT-registered, read its revenue trend, and describe its ownership footprint — all from official open data, returned as structured JSON. This is KYB (Know Your Business): the business-facing counterpart to KYC, and a natural fit for an agent.

What KYB checks in Lithuania

KYB verifies that a counterparty — a new customer, supplier, reseller, or partner — is a real, active, legitimate business before you commit to it. For a Lithuanian company, the core checks are:

• Existence and identity — a valid registration code (įmonės kodas) and matching legal name and form (UAB, MB, IĮ, AB).
• Legal status — active, versus in liquidation, bankruptcy, or deregistered.
• VAT registration — whether the company is a VAT payer, which matters for invoicing and tax treatment.
• Sector — the registered economic activity (EVRK), to sanity-check that the business does what it claims.
• Financial standing — revenue and profit trend, assets and equity, from filed statements.
• Ownership footprint — the makeup of participants (domestic vs. foreign, natural vs. legal).

The manual process today

Done by hand, KYB means visiting a registry portal, searching the name, confirming the code, opening the profile to read status and VAT, downloading financial statements to eyeball the trend, and noting the ownership structure. For one company it is tedious but doable. Across a supplier list or a batch of leads, it becomes a data-entry job that is slow, easy to get wrong, and impossible to keep current.

Automating it with an agent

Connected to ImonesMCP, an agent runs the same checklist through typed tools:

• Verify existence and status — search_companies resolves the name to a registration code; get_company returns the legal form and current status. If the status is anything other than active, that is a flag.
• Confirm VAT — the get_company profile carries VAT registration and code, and search_companies supports a vatActive filter for batch screening.
• Read the financial trend — get_company_financials returns yearly revenue, pre-tax and net profit, assets, and equity, so the agent can spot a declining top line or thin equity.
• Describe the ownership footprint — get_company_people returns aggregate counts of participants (Lithuanian/foreign, natural/legal). Useful for risk context; it never names individuals.
• Check the sector — resolve_activity_code translates the company's EVRK code into a readable section and division, so the agent can confirm the activity matches the relationship.

The company data guide explains each of these fields in depth.

Batch screening a list

The real payoff is volume. Instead of checking one counterparty, hand the agent a list of supplier names and ask it to screen them all. For each name it resolves the code with search_companies, pulls status and VAT with get_company, and returns a compact table — code, status, VAT active, latest revenue, sector — with the questionable rows flagged. You can pre-filter the population first: a single search_companies call with vatActive: true, a municipality, and a minRevenue floor narrows a market to the companies that meet your baseline before any per-company lookup. aggregate_companies complements this by counting how many companies match a profile by division, municipality, or status, which is useful for sizing a segment before you screen it.

A sample due-diligence prompt

Because the agent orchestrates the tools itself, your instruction can be plain language:

Run a KYB check on the Lithuanian company "Example Logistika UAB".
Confirm it is active and VAT-registered, summarize its revenue trend
over the last available years, state its registered sector, and
describe its ownership footprint (aggregate counts only). Flag
anything unusual. Cite the registration code you used.

A capable agent will call search_companies to get the code, get_company for status and VAT, get_company_financials for the trend, resolve_activity_code for the sector, and get_company_people for ownership — then return a short structured summary with the code it relied on. Start free and scale as your volume grows; see pricing for the plans.

Limits and what to verify manually

Automation gets you a fast, consistent first pass — not a legal opinion. Keep these limits in mind:

• No named individuals. Ownership is aggregate only. If your compliance process requires identifying beneficial owners, obtain that from the appropriate authorized register — it is not part of open data and never in ImonesMCP.
• Filings, not real time. Financial statements reflect what a company has filed, so the latest year may lag. Treat trends as indicative and check filing dates.
• Data is "as is." The source registers are authoritative, but no dataset is guaranteed complete for every record. Do not use it as the sole basis for a legal, credit, or compliance decision — corroborate high-stakes findings.
• Screening, not a verdict. The agent surfaces flags; a human should judge borderline cases.

Used this way, an ImonesMCP agent turns hours of manual KYB into a reliable first pass you can run on demand. Connect one from the docs.