Business intelligence · Data
Companero
Company intelligence for all of Romania
Year
2023 — prezent
Our role
Our own product — concept, architecture, development and production operations.
Status
in production
2.7M
companies indexed
5
official sources
MCP
AI tools for agents
The challenge
Data about Romanian companies is scattered across dozens of public sources — the tax authority, trade registry, Official Gazette, central bank, public procurement — each with its own format, cadence and traps. The challenge: bring them all into one place, always fresh, and make them searchable and useful in milliseconds.
The approach
We built a dual-database architecture (MySQL for relational, MongoDB for documents) with Elasticsearch for full-text search over 2.7M companies. Async importers grind through hundreds of thousands of records via message queues. On top of the data we layered an AI assistant with tool access (MCP) that answers complex questions, plus a suite of due-diligence tools.
From the product
What stands out
- Full-text search across 2.7M companies, indexed in seconds via streaming (~17k docs/s)
- Five import pipelines: tax authority, trade registry, Official Gazette, central bank, public procurement
- AI assistant with MCP tools — due diligence, people, tax debts, public contracts
- Person identity resolution: links directors and shareholders across companies
- REST API with JWT auth and credit-based billing
Tech stack
Next project
ElectrionERP