UMMINEX

Universal Market Matching Information Exchange

Be found without giving up control.

An open language for offers, requests, and the conditions under which they fit together.

UMMINEX makes Market Intents machine-readable so that different systems can understand what is offered, requested, and actually needed.

Offer and demand can exist side by side – and still fail to find each other.

A refrigerated truck is standing unused in a depot on Thursday evening.

Twenty-two kilometres away, someone is looking for exactly such a vehicle for that evening and would pay for it.

They may still never find each other.

Not necessarily because the information is absent. Both may even be public online. But they are not expressed in a common form in which a machine can reliably recognise: these fit together.

Offer and demand do not necessarily lack. A common language does.

Markets begin with intentions.

People and organisations do not simply look for “data”. They want to buy, sell, rent, let, book, offer, procure, deliver, exchange, work, commission, provide, or use something.

We call such a market intention a Market Intent. UMMINEX provides a shared machine-readable language for expressing these Market Intents and their conditions.

What are you looking for? What do you offer?

A Market Intent is more than an object. Relevant aspects can include:

Who?What?How much?When?Market IntentWhere?Under which conditions?In exchange for what?Which alternatives?What is mandatory?What is preferred?What is still unknown?

Not every detail needs to be known already. UMMINEX can treat missing or unresolved information as such. Unknown does not mean: no match.

Both sides need to fit together.

Offer

“800 aluminium housings per month available from March.”

Request

“Wanted: 800 to 1,000 aluminium housings per month from March.”

It is not always enough to check whether an offer satisfies a requester. The offerer’s conditions may also need to be met: minimum quantity, delivery area, period, payment method, admissibility, or availability.

UMMINEX therefore describes more than objects or listings. It describes Market Intents and the conditions under which different sides can be compatible.

A data record is not yet a market.

Product information

  • Aluminium housing
  • Type X
  • €24.80

Market Intent

  • 800–1,000 units per month
  • from March
  • delivery to Ulm
  • material requirements
  • minimum or partial quantities
  • possible alternatives
  • delivery capability and further conditions

UMMINEX does not describe things alone. It describes the conditions under which something fits together.

One principle. Very different markets.

Hotel

“Double room from 12 to 15 October, dog allowed.”

Doctor’s appointment

“Available appointment Tuesday between 9 and 12.”

Job

“Qualified care professional, at least 25 hours, no more than 20 kilometres away.”

Industry

“800 to 1,000 aluminium housings per month from March.”

Transport

“Two pallets from Lübeck to Munich by Friday.”

Exchange

“Machine use in exchange for maintenance work.”

The objects differ. The basic questions are similar: who wants what, when, where, in what quantity, under which conditions, and in exchange for what?

Why digital markets often need platforms today.

Digital platforms have solved a real problem. They collect information, standardise it, make it searchable, and bring market participants together. That is valuable intermediation.

A structural dependency can arise where discoverability is possible only when market information is maintained exclusively, or effectively necessarily, in one central system. This can mean repeated data entry, dependency on access rules and ranking, or a harder change of intermediary.

Intermediation is valuable. But the intermediary does not have to own the market.

UMMINEX is not intended to abolish intermediation. Search services, platforms, AI assistants, databases, directories, and specialised matching services can continue to offer valuable services. The shared layer of understanding should be open.

A common language instead of a common database.

Information can reside in different places and still follow a shared machine-readable structure. A hotel can publish availability, a company production capacity, a care home free places, or a practice available time windows.

Information should be maintainable where it originates. It should not have to be entered completely anew for every intermediary service.

Maintain once at the authoritative source – make it usable wherever it is needed.

Open data structures do not mean: no databases. Travel search services, procurement systems, directories, regional search services, AI providers, and specialised matching services can collect, index, organise, and build different services from accessible information.

UMMINEX does not abolish databases. It aims to allow databases to be services rather than necessarily the only gateway to a market.

An open standard for an open digital market

The technical problem behind UMMINEX has an economic consequence. To be found digitally today, people often maintain offers and availability where many requesters already search: on platforms, marketplaces, and specialised portals.

These services create real value. They collect information, make it comparable, and bring market participants together. A dependency can arise where discoverability and market access are inseparable from the data stock of one intermediary.

UMMINEX takes a different approach.

The intermediary does not have to control access to market data at the same time.

Data holder

maintains the authoritative information.

UMMINEX

describes meaning in a shared open form.

Index

can make information discoverable.

Matcher

can compare Market Intents and conditions.

AI assistant

can make complex search manageable for people.

Intermediary / platform

can provide convenience, advice, verification, trust, transactions, or other services.

None of these layers necessarily has to control all the others. Platforms, search services, and intermediaries can continue to offer valuable services and charge for them.

Intermediation remains valuable. Dependency does not have to.

Open standards change where market power arises.

A functioning and widely used UMMINEX could reduce structural dependencies on individual platforms. When offers and requests are published in an open form and processed by different services, possession of one central database is less likely to become the only gateway to a market.

UMMINEX does not dissolve network effects, brands, trust, reach, capital, ratings, computing capacity, or service quality. Open data structures are not a claim to eliminate market power.

UMMINEX vision: discoverability should not conceptually depend on one platform being the exclusive location of a participant's market information.

A building block for digital autonomy

Across different policy and legal frameworks, the European Union pursues goals such as interoperability, better usability of data, control for data holders, fair and contestable digital markets, and open digital ecosystems.

UMMINEX architecture and licensing principles have conceptual parallels with these goals. This is a UMMINEX interpretation of those goals – not EU endorsement, certification, or a claim that UMMINEX is an EU project or an implementation of the Digital Markets Act.

Digital sovereignty does not necessarily need the next gatekeeper.

One response to dependencies on large platforms could be to build comparable central platforms. Open standards allow a different thesis: many companies, public institutions, open-source projects, and commercial providers can compete on the same layer of understanding.

Sovereignty can also arise because fewer gatekeepers are needed.

UMMINEX is not a European protectionist standard. It is intended to be usable worldwide. Europe could benefit especially from such infrastructure because open standards, interoperability, data control, and open source are among goals formulated by EU institutions.

Built for the open Internet.

UMMINEX was originally developed by factsoft AG. But UMMINEX is not intended to remain proprietary technology of factsoft AG.

UMMINEX is publicly available under the Apache License 2.0. Separately licensed components remain separately identified. The license applying to UMMINEX and the public repository are available on GitHub.

Apache License 2.0

  • usable
  • modifiable
  • redistributable
  • usable commercially
  • express patent grant under the license terms
  • open further development possible

Not our standard. Our contribution.

We did not develop UMMINEX to operate the next central marketplace ourselves. In our view, a shared language for digital markets should be open infrastructure.

Others should be able to build on it, develop better implementations and commercial services, criticise UMMINEX, and develop it further. License conditions still apply.

The Apache License 2.0 includes conditions, in particular for license notices, modifications, and redistribution. Only the full license text is controlling.

UMMINEX – originally developed by factsoft AG, Munich, Germany, 2026.

View the Apache License 2.0 for UMMINEX

UMMINEX should not belong to us. It should work.

Open licensing is only the first step. An open standard becomes genuinely independent only when others can understand, implement, criticise, and develop it further.

Implement. Test. Disprove. Improve.

Available rooms in town.

Several hotels publish current availability at their respective sources. A tourist office or private travel service can collect this information and bring it together on its own website.

Available this evening

  • 3 single rooms
  • 7 double rooms
  • 2 family rooms
  • 1 accessible holiday apartment

Hotels do not necessarily have to maintain availability manually in that particular regional system as well. The authoritative information can be taken from their respective source. How individual services publish, find, update, or compare information remains a task of their respective implementations.

Make simple things understandable.

Is the museum open today?Does the tour take place despite the public holiday?Which doctor has an available appointment soon?Where is a care place available at short notice?Which workshop can repair my vehicle this week?Which supplier can deliver by Friday?

Information often already exists. The challenge is its precise meaning: for which time does it apply, which exception applies, which quantity is available, and which conditions belong to it?

For people, it remains a conversation.

“I am looking for a holiday job in Lübeck for my 17-year-old son, preferably in August, for exactly one week.”
  1. Human language
  2. Semantic understanding
  3. UMMINEX Market Intent
  4. Validation
  5. Understandable restatement

“I understand: a holiday job is sought for a 17-year-old in Lübeck. The work should take place in August and last exactly one week.”

People increasingly use systems to describe their goal in natural language. AI can support the translation between natural language and structured meaning. UMMINEX itself is not AI.

AI can infer meaning. UMMINEX makes meaning explicit and verifiable.

Email for markets?

Email works globally not because one company manages every message. Different providers and systems can communicate because they use shared open procedures.

UMMINEX follows a related principle for market information: different operators, different software, different databases, different storage locations – and one shared layer of understanding.

This is an analogy of architecture and competition. UMMINEX is not technically an email protocol.

What UMMINEX is not.

Not a platform

UMMINEX is not one individual intermediation interface.

Not a search engine

UMMINEX is not itself the search service.

Not AI

UMMINEX is an open layer of understanding, not AI.

Not a central database

UMMINEX does not prescribe one single intermediary.

UMMINEX is an open layer of understanding for Market Intents.

Open, precise, verifiable.

Market Intents

Offers and requests can be described together with their conditions.

Deterministic validation

Formal structures can be checked against defined rules.

UNKNOWN remains UNKNOWN

Missing information does not automatically become a negative statement.

Varied conditions

Time, quantities, roles, geography, consideration, alternatives, and resources can be structured.

An open layer of understanding.

Developers can use UMMINEX to create, validate, exchange, and further process structured market information. View source and specification on GitHub.

Who is behind UMMINEX?

UMMINEX was initiated and originally developed by factsoft ag in 2026. factsoft ag currently operates umminex.org and carries the organisational responsibility for this website and the current project phase.

UMMINEX is not intended to be a proprietary product whose use or further development permanently depends on factsoft ag. The published material is available under the Apache License 2.0 for factsoft-controlled project material.

Today

factsoft ag

Initiator, original development, website operation, and current organisational responsibility.

Already open

Apache License 2.0

The published material may be used, implemented, modified, and redistributed, subject to the license terms.

Next step

Community and governance

Independent implementations, critical review, counterexamples, and a long-term governance structure.

For an independent future

An open standard should not depend on a single company in the long term – not even on its original developer. UMMINEX should gradually move toward an open and independent governance structure.

Its long-term form has not been decided. Community governance, an independent technical body, an association, an organisation, a foundation, or another suitable open structure are possible models, not decisions already made.

UMMINEX was initiated by factsoft ag. The standard should be able to develop independently from individual companies over time.

Where UMMINEX stands today

UMMINEX is not a finished product. The semantic core of the standard is well developed and has been examined with systematic tests. Many difficult cases – from time and quantities to alternatives, resources, and multi-party relationships – can be formally represented and deterministically checked within defined test scopes.

That is not the same as proving that UMMINEX can already express every conceivable Market Intent completely and unambiguously. The next phase of the project begins at exactly that boundary.

The reviewed public release candidate is published on GitHub. The published state was verified again from a fresh clone after publication: the approved public test suite passed 18/18, and the canonical content matched the sealed RC1. This test evidence is not a universality proof.

Specified

What is already defined

The formal core covers, among other things, actors and roles, offers and requests, quantities, time, geography, admissibility, consideration, preferences, alternatives, resources, multiple parties, evidence, UNKNOWN, conditions, matching, and global feasibility. These are examples, not a complete normative list.

Strongly tested

What tests have shown

Test families examine individual constructs, combinations, Validator, Builder, and Matcher behavior, resources, alternatives, time, multiple parties, cross-root relationships, evidence resolution, UNKNOWN, and controlled roundtrips. Found errors led to reproducible repairs and regression tests.

Still open

What needs more work

Broader encoder coverage, some Matcher handoffs, natural language, independent implementations, real-world pilots, discovery, indexing, updates, and scaling need further evidence and practical experience.

UMMINEX separates the description of meaning from its later evaluation. Structured Market Intents can be formally validated; matching is intended to rely on explicit conditions. UNKNOWN ≠ NO_MATCH.

Governance

UMMINEX already has technical governance rules for Authority, versioning, and changes. Long-term independent institutional governance remains an open project task. Decision-making, evaluation of changes, maintainer structure, different implementations, conflict resolution, succession, institutional stewardship, and protection of the standard's openness still need to be developed.

What we can say – and what we cannot.

What evidence supports

The semantic core, validation, and matching have been examined systematically and with counterexamples in defined scopes. Numerous different constructs have been tested not only in isolation but also in combination.

Very strong test evidence is not a universality proof.

What has not been proven

There is no mathematical proof that every conceivable Market Intent can be represented. Nor is it proven that arbitrary natural language can generally be translated into UMMINEX without loss and reconstructed exactly.

Perfect AI interpretation, complete worldwide discovery, and a finished global ecosystem are not claimed either.

The most difficult interface is the human one

People formulate incompletely, contextually, and ambiguously. Practical systems must identify what is certainly meant, what follows from context, which interpretations differ materially, and when a question is justified.

Encoders, controlled test systems, and extensive experiments exist. General translation of arbitrary natural language into UMMINEX is not yet solved reliably. Technical roundtrips in controlled settings are not automatically natural-language semantic roundtrips.

From a fitting pair to an open ecosystem

Discovery and publication

Publication and Discovery have specified foundations. A worldwide discovery, indexing, and update ecosystem has not yet been built. Practical experiments with publication, discovery, caching, change detection, distribution, and scaling are needed.

Vocabularies

Different technical terms, classifications, and schemas need to be related. Further work includes equivalence, similarity, broader and narrower terms, mapping uncertainty, and domain testing.

Matching and solvers

Multiple parties, scarce resources, alternatives, and dependent conditions can concern an entire possible allocation. Concepts and implementations exist for these cases and need more counterexamples and application scenarios.

Real-world pilots

Artificial test cases are not enough. Pilot projects are needed in commerce, industry, services, tourism, appointments, jobs, care, transport, municipal offers, and capacity allocation.

1 Semantic core
well developed
2 Validator, matching, and tests
well developed, ongoing
3 Usable encoder
in development, not finished
4 AI encoder
experimental, not approved
5 Decoder: UMMINEX → understandable language
controlled restatements, more work needed
6 Independent implementations
next phase
7 Real-world pilots
beginning
8 Discovery and indexes
to be built

UMMINEX needs collaborators now.

An open standard cannot be defined, tested, and interpreted by a single development team indefinitely. We are looking for people and organisations who will try to disprove UMMINEX, implement it, and test it against real problems.

Test semantics

Bring real Market Intents and, especially, valuable counterexamples.

Implement

Develop independent Validators, Encoders, Decoders, Matchers, Resolvers, indexes, or tools.

Test languages and roundtrips

Examine text → UMMINEX → text, including additional languages.

Build discovery

Experiment with publication, discovery, indexing, caching, and updates.

Attack matching

Search for cases where MATCH, NO_MATCH, or UNKNOWN could be treated incorrectly.

Document and connect

Improve examples, translations, tutorials, and interoperability with existing systems.

The most valuable case is the one UMMINEX cannot handle yet.

Finding a realistic Market Intent that cannot be expressed correctly with current semantics helps the project. Counterexamples, edge cases, ambiguities, contradictions, unexpected combinations, and implementation differences are explicitly welcome.

Help build it. Or try to disprove UMMINEX.

The public repository provides source and specification for independent inspection. For contact: contact@umminex.org.

The market does not need a single owner.

UMMINEX does not create a new central marketplace. It creates a shared language on which different providers, requesters, search services, databases, platforms, and AI assistants can build.

Intermediaries can continue to intermediate. Databases can continue to collect data. Platforms can continue to offer convenient services. But the shared layer of understanding can be open.

Be found without giving up control.