Pricing philosophy, product launches, and the engineering choices behind the LexAPI corpus. Filed by whoever shipped the thing.
A model drafts the passage a relevant judgment would contain, and retrieval runs on that. Nine landmark-case queries against production: first-place hits nearly doubled, MRR +36% — with graphs, latency numbers, and the two queries where HyDE loses.
Read onTwo lines to your first parsed regulation. Hand-written thin clients — retries that respect Retry-After, typed errors you can branch on, credit visibility on every response, streaming export, and point-in-time retrieval.
Read onA reproducible walkthrough on a hard GDPR / AI Act question — what semantic retrieval and a real typed citation graph do that keyword search on EUR-Lex simply can’t.
Read onInstall one npm package, paste your API key into your MCP config, and Claude (or Cursor, or any MCP-enabled client) can search EUR-Lex, fetch documents, and traverse the citation graph natively.
Read onA monthly pool of credits instead of daily caps. What each operation costs, what the Free tier now includes, and why existing customers keep their original plan.
Read on