Pre-launch preview. Pauhu® Ltd is building toward launch. Explore the foundation and register your interest; subscriptions are not open yet.
Pauhu

Welcome data

Welcome. You are five minutes from your first sourced answer.

This page walks you from zero to a working query against Europe's sourced foundation. Three steps to your first answer. Two more to integrate it into your own stack. The free tier opens the foundation to you on day one.

1

Pick your path.

Every path reaches the whole corpus. Pick the one that matches how you already work.

  1. Connect (MCP)You want your AI assistant to answer with European sources, or an honest gap. Best for analysts, lawyers, policy leads, procurement officers, anyone working in a chat client today.
  2. Build (REST)You want sourced answers inside your own code. Best for engineering teams, data platforms, compliance automation.
  3. Stay current (stream)You want continuous ingest, the moment Europe publishes. Best for monitoring systems, alerting pipelines, data warehouses.

Pick one to start. Pauhu® keeps the other three ready for the day you want them.

2

Get your key.

Sign in with your professional e-ID, or with email. The free tier activates immediately: fourteen days of full Pauhu® MCP, every domain, every language. Payment becomes a question only when you decide to keep going.

3

Your first sourced answer.

Ask anything within the Common European Data Spaces foundation. From your AI assistant, ask: What does Article 5 of the AI Act require, with sources?

Pauhu® returns the sourced paragraphs, the CELEX identifier, the source URL, and the recital that backs the operative provision. Click any source to verify it on EUR-Lex. For a question within a still-filling domain, the response is an honest gap that names what would close it. Both behaviours are the product.

From your terminal (Build)

curl -H 'Authorization: Bearer YOUR_KEY' \
  'https://api.pauhu.eu/v1/query?q=AI+Act+Article+5+purpose+limitation&lang=en'

The response carries the answer text, the cited rows with source URLs, and a response shape your code can route by: answer, honest gap, or named conclusion. Three shapes, exhaustive, one always applies.

4

The enrichment layers.

Every sourced row carries structured metadata. Use the citation chain on its own, or pull the metadata for downstream work where you want your output to be definitive.

{
  "celex": "32024R1689",
  "title": "Regulation (EU) 2024/1689 (AI Act)",
  "domains": ["public-administration", "research-innovation"],
  "obligations": 147,
  "prohibitions": 23,
  "permissions": 89,
  "exemptions": 12,
  "cpv_codes": ["72000000", "48000000"],
  "languages": ["en", "fr", "de", "fi", "sv", "..."],
  "cross_references": ["32016R0679", "32022R0868"],
  "terminology_entries": 342
}

Every field above comes back with its European source, or an honest gap that names what would fill it. The metadata is the product as much as the prose answer is.

5

Build on it.

What practitioners build on the foundation:

Compliance workflows

Every answer carries the source paragraph that backs it, or an honest gap when the foundation is still filling.

Regulatory change detection

Twenty-four European languages, the moment Europe publishes.

Procurement intelligence

TED notices combined with the legal substrate that governs them.

National implementation maps

European instruments cross-referenced with member-state transpositions.

Resources

Resources.

Close

The foundation is here. Build.

Questions: hello@pauhu.eu. Helsinki, EU jurisdiction, EU timezone.