The idea behind Prexfo #

Prexfo is more than a technical format. It is a deliberate choice against the prevailing trend of moving data into central, API-driven platforms — and for the owner’s control over their most valuable digital asset. Three pillars carry this idea.

Pillar 1 — Data sovereignty #

In the classic world, truth lives in a vendor’s database. Whoever switches vendors loses data, history and negotiating power. Prexfo instead defines the combination of PropertyExchangeFolder and PropertyExchangeFormat as the canonical source of truth. Any database a program builds from it is explicitly considered a subordinate, at-any-time replaceable cache.

The consequence is unmatched resilience: a company can lose its entire software infrastructure and restore it completely from the Prexfo archives alone. That drastically lowers the barrier to switching vendors and strengthens the data owner’s position.

Pillar 2 — Decentralization #

No central server, no open API that exposes the whole database at any moment. Whoever wants to share data deliberately creates an export for a specific purpose and a specific recipient. This enforces a need-to-know decision — and makes uncontrolled access, as API-based systems allow, technically impossible.

A statement, not just technology. Where platform vendors build ecosystems to lock customers in, Prexfo is the counter-model: it deliberately addresses those who fear big-tech dominance and loss of control.

Pillar 3 — Lifecycle #

Transaction formats like OpenImmo capture only a brief moment: marketing. Prexfo captures the whole life of a property. The append-only model attaches every event — planning, construction, maintenance, tenant change, modernization, sale — as an immutable record. The result is a complete, auditable history that is worth its weight in gold for due diligence, risk assessment and valuation.

AI as a tool — not a data sink #

Prexfo rejects the common pattern of dumping sensitive company data into the black box of third-party AI models. Instead: AI generates tools, not answers from raw data.

A bank might ask the AI: “Write me a script that extracts all maintenance invoices from a Prexfo structure over the last five years, groups them by trade and calculates the cost per square meter.” The AI delivers the script. The data owner runs it on a controlled export. The raw data never leaves the secure environment.

This follows the UNIX philosophy: small, specialized tools operating on a standardized data format. And it is the compelling argument for security-sensitive actors — banks, insurers, public institutions: harness the productivity of AI without sacrificing data sovereignty.

The difference in one sentence. Not “send your data to the AI”, but “have the AI build a tool that you apply to your data”.