Cloem® creates, then timestamps and optionally publishes variants of your patent claims, called cloems™, which may be prior art (check with your patent attorney).

- Cloem combines proven human claim drafting expertise and computational linguistics to create cloems. Cloems are technical poems ("poem" + "claim" = "cloem"). We leverage the best dictionaries (like Wordnet, Wikipedia and other proprietary lexicons derived from 70 millions patent documents).
- each cloem is associated with a unique and permanent address and with a timestamp (mathematical token used to establish a reliable date of creation).
- cloems can be optionally published ("public cloems") or can remain unpublished ("private cloems"). Public cloems are accessible to anyone.
- depending on your jurisdiction (consult a patent attorney), cloems may constitute enabling prior art. In other words, it can be that a particular cloem can destroy the novelty i.e. anticipate later filed patent applications. Alternatively, one or more cloems can endanger the inventive step of subsequent patent applications. Prior art or not, cloems can inspire you. For example, private cloems of yours can help you to draft better claims. Read the "Benefits" section for more use cases.

In more details...


From patent claims (independent claims or entire claim trees), that we call “base claims”, we create numerous textual variations thereof. These variants are technical poems and are called “cloems”.

These patent claims are provided or selected by you. For example you can use base claims which are already published (for example those of your own patent application after 18 months or a set of claims of a competitor). If your base claims are not yet published, we will not disclose them if you want to keep them confidential and/or associated cloems (contact us if you need a NDA, although this is not our current policy). Please note that we do not draft nor select base claims for you in our existing commercial offer.

Part of our model, we have algorithmically "encoded" the best claim drafting techniques, i.e. human expertise, gained from our experience at major companies (respectively world leaders in software and in healthcare) and at law firms ("private practice"). The art of drafting follows - at best - the practices observed in industry, and also follows the EPO Examination Guidelines and the USPTO M.P.E.P. Numerous sources have been analysed and incorporated, for example the Landis drafting handbook, claim construction dictionaries, TRIZ, litigated patents, patent standards and many more. We use specialized a combination of modified NLP parsers, specifically designed to handle patent claims which present numerous grammatical challenges. We combine with this know-how and grammar analysis with a wide range of dictionaries and ontologies, which we also have processed to fit patent contents. Among many other sources, we use Wordnet (e.g. for the thesaurus, providing for example synonyms, hyponyms and hyperonyms) and Wikipedia (e.g. for "fresh terminology"). Of course, we also intensively studied the patent corpus (over 70’000’000 patent documents have been analysed). For example, we have now unique and exclusive dictionaries with IPC subclass granularity. In fine, we also perform crowdsourcing, human supervision, text mining, semantic techniques (and other undisclosed techniques) along the creation process.

As a result, cloems are (tentatively) grammatical (i.e. of verified syntax and orthograph). Cloems are both machine and human creations (contributions are deeply intermingled).

Permanent URL per Cloem

After its creation, each and every cloem comes with a unique and persistent web address.

Corresponding QR codes are also provided: you can share any cloem.


We use trusted timestamping so that a given cloem can be associated with a reliable date.

Even if not required by most jurisdictions in the world, we are striving to get the highest level of mathematical proof available. We mostly use RFC 3161 for now. We are using other timestamping techniques as well.



Source: Wikipedia

Optional publication

Publication is always optional.

You decide if - and when - you want to publish.

a) “Public cloems”: you can decide to publish cloems. They then become searchable and available to the public. You can choose the publication date of cloems. You can decide to publish immediately after the creation of your cloems or at a later date of your choice (for example 18 months after your own corresponding patent filing).  You can choose to publish the original patent claims, or not. You can choose an “assignee name” and choose whether to publish it along your cloems, or not. Beware: once published, one or some or all of cloems cannot be removed from public availability.

b) “Private cloems”: alternatively, you can decide to keep cloems private and confidential. In this case, you - and only you - can search in cloems associated with your patent claims (cloems are not disclosed to the public). You can thus study these cloems and file regular patent applications based on one or more of them. Even with published cloems, grace periods provisions may apply depending on your jurisdiction.

Please note: publication is irrevocable. Once published, we will not remove one or more cloems (unless laws require it). Again, we highly recommend you to consult a patent attorney.



"Public" cloems are published. They are searchable. Public cloems are "made available to the public".

"Private" cloems are not published: they are technically protected and they are visible only to those of you with appropriate access rights. They are not part of the public index.

Your search queries are not public, they are protected with the highest level of SSL encryption possible today (SSL Extended Validation). This default feature tries to defeat eavesdropping of your queries. We do not publish search queries. 

We are permanently striving to provide the best search, browsing and visualization features.

Enabling prior art ?


Some cloems can be enabling disclosures and may be eligible to the status of prior art.

Any enabled publication which is accessible to the public is normally considered as prior art. A cloem derived from a patent claim is as enabled as any given text, chosen randomly. Not more but not less. Some destroy the novelty of subsequent patent applications, many attack the inventive step of such attempts. Some make sense and contain a real technical teaching. Some do not, but many contain incentives to real human inventions or further claimed subject matter. The common general knowledge - which is increasing - enables the skilled person to carry out the cloem.

Cloems are thus very ordinary texts. The enablement power of cloems are thus exactly the same as any ordinary text: it is to be decided on a case by case basis whether the considered text is an enabling disclosure or not. Cloems probably can be compared to paragraphs present in patent descriptions (which are also prior art). Patent attorneys often first write patent claims and then provide support for these claims by detailing and adding support to these claims. Patent claims must be read in combination with associated descriptions. Borders are nevertheless fuzzy.

Patent claims, boundaries of protection

Patent claims represent the legal boundaries of protection. Yet these claims, as natural language, reveal language ambiguities. Cloem is a remediation to these fuzzy protection boundaries. With cloems, we handle a massive “linearisation” of such fuzzy and ambiguous texts by creating clear and litteral versions.