Public procurement, funding calls and company registers from SK, CZ and the EU — normalized into one schema, refreshed daily, with per-record provenance. Available via MCP, REST API or bulk files.
One record, end to end
The last two fields ship on every row of every dataset. That is what you are actually buying.
Public data in CEE exists — but it’s spread across dozens of portals, in different formats, with different entity IDs, and most of it has no usable API. Building and maintaining that pipeline yourself takes weeks and then breaks silently. We keep it clean, joined on registry IDs, and current.
Every card carries the same fields so you can compare them. Planned means not built yet — whoever asks first sets the priority and the schema.
Notices, awards and contract values, normalized to one eForms-compatible schema
Small-value contracts that never reach TED, collected from national and municipal portals
Open calls, deadlines, eligibility and budgets across national and EU programmes
Who won what, at what value, and how their win rate develops over time
Who buys what, how often and from whom — aggregated per contracting authority
Identifiers, financial statements and officers, joined on registry IDs
Who owns whom, including ultimate beneficial owners, as a traversable graph
The same tender schema extended to neighbouring markets
Samples are 100 rows, CSV and JSON, no e-mail required. Volume, freshness and API keys are behind the request form.
Three delivery modes, same schema. Not sure which fits? Tell us your stack in the form and we’ll set it up.
One MCP server exposes every dataset as tools your agent can query directly — no glue code.
{
"mcpServers": {
"delphinium": {
"command": "npx",
"args": ["-y", "@delphinium/mcp"],
"env": { "DELPHINIUM_API_KEY": "dlp_live_..." }
}
}
}Every extraction is schema-validated, deduplicated on registry IDs (IČO/IČ), and a random 1% sample is re-verified against the source each week. Pipeline status and last-run times are public.
Feed your agent or app with clean CEE data instead of building 20 scrapers. MCP + API, usage-based.
Search, monitor and export tenders and grants without watching 40 portals. Saved searches and alerts.
What we’re considering next. Tell us what you need and it moves up.
Indicative — we’re still calibrating. Building something and the price doesn’t fit? Tell us: early partners get pilot terms.
We reply within 24 hours with a sample and a schema for your use case.