Infrastructure for public knowledge
01 — The problem
Every year, thousands of institutions produce analyses on public reality — poverty, healthcare, structural funds, public spending. Each one is an act of understanding.
Each one disappears. A PDF on a website, cited for a few months, then gone. They do not connect. They do not build on each other. The AI agents analysts use every day have no memory of any of this — every session starts from zero, every institution starts from scratch.
The knowledge exists, dispersed across thousands of documents and hundreds of open data portals. But it is invisible to the tools that need it most — and none of it is connected.
02 — How it works
Connect once via MCP. Your AI tool queries the graph before every analysis — and publishes results after. Knowledge accumulates as a side effect of work already being done.
Layer 1 — Open data
The world's public data,
structured as a graph.
Entities and relations from governments, statistical agencies, and international bodies — PNRR, ISTAT, Eurostat, World Bank, OECD. Territories, institutions, spending flows, socioeconomic indicators, policies. Not files on portals: a connected fabric of real-world entities.
Layer 2 — Analyses
Every finding anchored
to the data it used.
A directed acyclic graph of institutional analyses, each linked to the open data entities it studied. Relationships between analyses are explicit — extends, contradicts, corrects. The history of understanding on any topic, navigable and cumulative.
Connect your AI tool
One MCP configuration. Works with Claude, ChatGPT, and any MCP-compatible agent. Your institution's analyses are attributed from your API key — no extra steps.
Query context before you start
Before any analysis, the agent calls search_analyses and query_entities. It retrieves prior analyses on the same topic and pulls structured data on the relevant entities — municipalities, funds, indicators. It does not start from zero.
Publish the result automatically
When done, publish_analysis adds a new node with explicit relationships: extends, contradicts, corrects. The graph grows. The next analyst inherits everything.
03 — The vision
"Quel che videro i miei occhi fu simultaneo; ciò che trascrivo, successivo — perché tale è il linguaggio."
Borges — L'Aleph, 1949
Knowledge is simultaneous. Language is sequential. Every tool we have for understanding the world betrays it. Aleph is built from the conviction that public knowledge deserves better infrastructure.
Read the essay →04 — Early access
We are onboarding early institutions and research teams for the private beta. Leave your email — we will reach out when access opens.
No spam. Contacted individually when beta opens.