The problem

Most patent value is lost before an examiner ever sees the application.

Not because the work was bad. Because the decisions that determine whether a patent is worth anything are all fixed at the moment of filing: what was captured, how it was framed, and what was disclosed before anyone checked.

By the time any of it surfaces, in prosecution, in diligence, or in a letter with a deadline, it is expensive and usually beyond repair.

What it costs

Every expensive outcome starts as a cheap decision.

Read each row left to right. The decision on the left costs almost nothing to get right at the time. The consequence on the right is what it costs to find out later.

  1. Describing the outcome instead of the mechanism

    Examination

    Rejections on eligibility, or a valid claim nobody needs to infringe

  2. Showing the work before checking the clock

    Foreign filing

    Rights barred outright in absolute-novelty countries. They do not come back

  3. Describing one prototype, claiming a genus

    Prosecution, then litigation

    Claims narrowed to the prototype, because the support was never written

  4. Assigning inventorship by seniority

    Diligence

    One broken assignment link can make a strong portfolio worthless to a buyer

  5. Claiming the product rather than the principle

    The competitor's release

    A patent that is valid and routed around, or that cannot be proven infringed

  6. Searching after drafting, when nobody wants to redesign

    The PTAB, or an assertion letter

    70% of patents reaching a final written decision lost all claims in FY2024

This is the whole argument for working early. Upstream, these are conversations. Downstream, they are write-offs.

Where it happens

Ten places the ground gets given away.

  1. 01

    Ideation

    New-to-us gets mistaken for new-to-the-world, and the defensible invention hides in calibration, orchestration or the manufacturing sequence rather than in the headline result.

    We work from the mechanism rather than the outcome, and test candidates against the published landscape before anyone spends on them.

  2. 02

    Capture and evidence

    Records get reconstructed months later. Baselines are missing, prophetic and actual examples blur together, and the human reasoning behind an AI-assisted result is gone.

    A contemporaneous technical record built while the work happens, which is what later inventorship and support arguments are made of.

  3. 03

    Prior-art search

    Keyword-only searching, in one language, against terminology engineers do not use. Nobody assembles the obviousness combinations.

    Classification-led search across languages and non-patent literature, with the scope, date and known limits stated on the face of it.

    Clearance and validity opinions go to a practitioner.

  4. 04

    Disclosure and priority

    A conference paper, a pitch deck, a pilot, a public repository. In absolute-novelty countries a disclosure before filing is not an argument to be won, it is a bar.

    A disclosure inventory and a jurisdictional calendar, so the question is answered before the talk is given rather than after.

    The highest-severity failure of the ten. Rights lost here do not come back.

  5. 05

    Inventorship and ownership

    Inventorship assigned by seniority and never revisited when the claims change. Contractor and joint-development ownership left unresolved.

    We establish the factual record, claim by claim, that the determination has to rest on.

    Every determination goes to a practitioner.

  6. 06

    Specification support

    The specification is written from what was built rather than from what has to be defensible. Ranges, endpoints and fallback positions are simply absent.

    A support matrix tying each intended claim feature to text, drawings, examples and a priority date, built before drafting rather than discovered after.

  7. 07

    Claim architecture

    Unnecessary vendors and sequences written into the independent claim. No ladder from broad to narrow, and no plan for proving infringement.

    Claim architecture designed for survival and competitive effect rather than for the path of least resistance to allowance.

    Drafting and prosecution go to a practitioner.

  8. 08

    Jurisdiction

    One filing assumed to produce worldwide rights, and countries chosen by where sales are today rather than where competitors manufacture.

    A country-value model built on where the position actually needs to bite, with the cost of each option stated.

  9. 09

    Portfolio and diligence

    Patent count used as a proxy for strength, annuities paid on assets with no role, and a data room holding application numbers instead of claim charts.

    Claims mapped to products, revenue and competitors, with keep-or-prune criteria you can act on.

  10. 10

    Assertion defence

    The first serious claim chart gets built only after the letter arrives, and engineering starts redesigning before anyone knows which elements matter.

    The position mapped and costed against what the product has to keep doing, handed to your counsel as something they can act on.

    Opinions and litigation strategy go to a practitioner.

The boundary

Why we can tell you not to file.

Kenorai is not a law firm and issues no legal opinions. We are also not paid to prepare filings, which is the part that matters: a firm that bills for filings cannot credibly be hired to attack the case for filing.

Where the work reaches a legal determination it stops and hands your practitioner the question, with the technical record already built.

Start

Find out early, while it is still cheap.

Tell us what you are building and where you think the gaps are. We will tell you whether you are right.

Start a conversation