Albert Construe is the claim interpretation module of Albert, Evolve’s proprietary AI agent for patent validity analysis. It construes a set of claims, sets out with supporting evidence what each unclear term means, highlights any inconsistencies caused by other disclosures within an application, and flags where amendments may have changed the scope of a claim term.
Impacts of G 1/24, G 1/25 and G 1/26 on claim interpretation
The recent decisions in G 1/24 and G 1/25, and the pending referral G 1/26, relate to description amendments and interpretation of claimed subject matter in view of the description and drawings. We have previously discussed these referrals (Evolve Insights) and their impacts. As a brief recap:
- G 1/24 established that the description and drawings must always be consulted when interpreting the claims (Evolve Insights).
- G 1/25 established that the description must be amended when an inconsistency between the claims and description leads to non-compliance with one or more requirements of the EPC (Evolve Insights).
- G 1/26 raises the question of how the description should be used to interpret the claims and what the impact will be (Evolve Insights). The Enlarged Board of Appeal has not issued an opinion on this referral yet.
What is clear from these referrals is that the disclosure within an application’s description or drawings can alter the meaning of a claimed term, and, consequently, the scope of protection conferred.
The practical impact of G 1/24 and G 1/25 is already being felt. In T 0977/25 (Evolve Insights), the Technical Board of Appeal applied both decisions in interpreting the claims, and in T 1849/23 (Evolve Insights) G 1/24 was used to broaden otherwise clear claim language. The Boards have also confirmed that the claims can be construed differently in the application as filed and in the granted patent (Evolve Insights), and that description amendments can extend protection (Evolve Insights).
An altered claim scope can have substantial consequences for the patentability of an application or the validity of a granted patent. The addition, removal or alteration of a definition that is interpreted as broadening the claims could introduce added matter or, post-grant, invalidate the patent. Equally, amendments that unintentionally narrow how a claim is construed may affect whether your product, or a competitor’s, falls within the scope of the claims. The practical drafting lessons are set out in Beware of boilerplate (Evolve Insights).
We have designed Albert Construe specifically to identify such issues.
What is Albert Construe and how does it work?
Albert Construe is a proprietary AI tool that we have developed at Evolve. It is built upon a curated knowledge bank of EPO case law decisions and the expertise of our own patent attorneys. The tool conducts an in-depth analysis of the claim language in the context of the specification as a whole, quickly identifies clarity issues and efficiently flags potentially problematic inconsistencies for further attorney review, including flagging risks associated with adaptation of the description. At Evolve, our existing integrated and adaptable AI systems allow us to incorporate Albert Construe as part of our day-to-day prosecution and opposition workflows, providing us with an additional and efficient patent validity check.
For each claim term, Albert Construe sets out its ordinary meaning, its meaning in the claims, the meaning given by any definition in the description, and the meaning that emerges from the examples, and then flags where those readings do not agree. Where we have both the application as filed and the granted or amended text, it also lists every difference between the two and reports how each might affect the interpretation of the claims.
Albert Construe is particularly useful for reviewing proposed description amendments and flagging any risk that the amendments could change the scope of a claim term. It is also where the other Albert modules start: any novelty, inventive step, basis or support analysis can be preceded by Albert Construe, which fixes what each feature of the claim means before the analysis is run on it, so that a clarity attack, or a dodgy definition, does not catch us out later.
How we use Albert Construe at Evolve
We use Albert Construe to assist us in drafting and prosecuting patent applications, and opposing patents, efficiently. Crucially, our attorneys make the final decision on how to define or interpret a claimed term and whether or not to make amendments to the description. Albert Construe helps us to be fully informed when making that decision.
As patent attorneys, we understand the utmost importance of confidentiality and protecting client data (Evolve Guides). This means no storing of client data, no training on inputs and enterprise-grade security. Our AI systems are built directly using foundational LLMs (Evolve Insights) from Google and Anthropic and not with any third-party software providers or SaaS companies. This gives us the flexibility to adapt our AI systems to changes in patent law, the science and the capabilities of the underlying LLMs.
How Albert Construe is built
Every Evolve agent is built from the same four parts. The case context: the client’s files, the prosecution history, the related family members and the instructions, connected from the case project folder. The legal context: the statute and case law of the jurisdiction, the tests the Boards and the courts actually apply, and our reading of where that is heading, kept current as decisions are issued. The instructions: the rules the agent follows, written by our own attorneys, which is where our expertise lives. And the workflow: the steps the agent takes with the attorney, interactive at every one, so that the attorney can challenge a finding and have it re-run.
Any of our agents can be called by name by any Evolve attorney or implementor, in the Claude Cowork desktop app or from inside Word, with a plain instruction such as “Run Albert Construe on this file”. The agents are connected to the relevant case folders and legal resources and can call on each other, and on our coding scripts, as the job needs. We hold our AI to five tests: it adds tangible value and solves a problem we actually have, it runs on the latest foundational models, it works within the files and software we already use (Word, Excel and SharePoint), it carries our own expertise rather than someone else’s, and it is adaptable, transparent and usable by everyone in the firm. Our Evolve AI page sets out the thinking behind this.
A complete AI ecosystem for pharma IP
Albert Construe is part of Albert, our suite of AI modules for patent validity analysis, and of Evolve’s wider portfolio of bespoke AI agents built specifically for pharmaceutical and biotech IP. Our clients benefit from the full ecosystem:
- Albert: Validity and opposition. Novelty, inventive step, support, added matter, priority and data strength, laid out for us to review. Albert Construe adds claim interpretation to that toolkit.
- Popper: Prosecution. Translates and summarises the office action and sets up the groundwork for our response.
- Darwin: Drafting. Claim sets, amendments, specification sections and ST.26 sequence listings, prepared to our instructions.
- Mendel: Amendments. Claim amendments and basis tables for EP, US and other jurisdictions.
- Gilbert: Quality. A second read of specifications, sequences, compounds and responses, with each issue graded.
- Franklin: Invention spotting. Finds inventions and disclosure risks in draft publications and competitor filings.
- Cassandra: Due diligence. Gathers and organises the evidence on coverage, validity, ownership and contract terms.
- Ammonite Chart: Patent estate and loss-of-exclusivity charts that show how long the protection runs.
FAQs
What is claim construction, and why does it matter now?
Claim construction is working out what the words of a claim actually mean, and therefore what the patent covers. Since G 1/24, the EPO must always consult the description and drawings when doing so, which means a definition, an example or a stray sentence in the description can widen or narrow a claim without the claim itself changing. That affects validity, infringement and the risk of added matter.
When should Albert Construe be run?
Before filing, to catch definitions and boilerplate that could later change the scope of the claims. During prosecution, whenever description amendments are proposed under G 1/25. And at the start of an opposition or a validity review, so that the meaning of every term is settled before the novelty and inventive step questions are asked.
Does Albert Construe decide what a claim means?
No. It sets out the candidate meanings and the evidence for each, and flags where the claims, the definitions and the examples do not agree. Our attorneys decide how to define or interpret the term and whether to amend. The output is there to make that decision fully informed.
Does it work for US applications as well as EP?
The knowledge bank is built on EPO case law, so the clarity and description-adaptation analysis is European. The claim-by-claim comparison of the application as filed against the granted text, and the flagging of definitions that shift claim scope, are useful for any jurisdiction, and we use them alongside Mendel when preparing US and other national claim sets.
Is our confidential information safe?
Yes. Albert Construe runs on foundational models from Google and Anthropic under enterprise terms, with no storing of client data, no training on inputs and no third-party software layer between your documents and the model.
Working with Evolve
If you would like to hear more about how we are using proprietary AI tools to assist in our work, or how Evolve can help you with your IP, please contact one of our attorneys. You can also read the original launch article for Albert Construe (Evolve Insights).