The Invention Disclosure Is the Patent: How Engineering Teams Get Better Applications for the Same Money
By Brandon Furdock, Registered Patent Attorney, Chicago Patent Law LLC
Here’s a trade secret from the drafting side of patent practice: two companies can pay the same fee, to the same attorney, for inventions of the same quality, and receive patents of wildly different strength. The variable isn’t the law or the budget. It’s what happened in the first meeting, when the invention moved from the engineer’s head into the attorney’s.
Patent quality is manufactured at disclosure. Everything after is finishing work.
What your attorney actually needs (it’s not what most teams provide)
The typical invention disclosure is either a starved paragraph (“we built a faster way to do X”) or, increasingly, its opposite: a sprawling AI-generated document that describes everything and distinguishes nothing. Both fail the same way, because they don’t answer the questions claims are built from. Those questions are few:
What was broken, technically, in the best existing approach? Not the market gap, but the engineering deficiency. “Prior retrieval systems re-embed the full corpus on every update, which costs O(n) per write” is the sentence claims are built on. “Search was slow” is not.
What’s the mechanism of your fix? The specific architectural or algorithmic move that produces the improvement. It’s the thing that, if a competitor copied it, would be the copying. Attorneys can dress a mechanism in claim language; they cannot conjure one from a description of outcomes.
Why isn’t it obvious? What did you try that failed? What does the field believe that your approach contradicts? Dead ends and surprises are prosecution gold. They’re the story that defeats an obviousness rejection, and they almost never appear in disclosures unless someone asks.
What are the variations? The other ways to implement it, the parameters that could change, the adjacent use cases. These become dependent claims and design-around blockers. Engineers know them instinctively and write them down never.
How would you detect infringement? If a competitor used this, what would be visible in their product, their docs, their papers? The answer shapes which version of the invention deserves claims at all.
The one-hour habit that changes everything
Teams that consistently produce strong portfolios all run some version of the same routine: a live disclosure session with the inventors, patent counsel, a whiteboard, and an hour, instead of documents lobbed over a wall. The economics are lopsided in its favor. An hour of structured conversation routinely surfaces the failed experiments, the variations, and the “wait, that part’s novel too?” moments that written disclosures omit. It’s where one invention becomes three, or where a weak filing is correctly and cheaply killed before drafting. It is the highest-leverage hour in the entire patent process, and it’s the one most companies skip to save time.
A note on AI-drafted disclosures, since they’re now everywhere. Used well, with the engineer dictating raw material, AI organizing it, and the engineer correcting it, they’re a genuine improvement over the starved paragraph. Used lazily, they’re worse, because they bury the three sentences that matter under twenty pages of confident filler, and someone still has to find them. The disclosure session is how. And however the disclosure was produced, it should record which AI tools contributed to the invention itself. Inventorship documentation is now part of basic hygiene for AI-native teams.
What to expect back (and what to push back on)
Disclosure quality also sharpens the client’s side of quality control. When you review a draft application, you’re checking three things, in order. Is the technical description accurate and complete, meaning you would hand it to a new hire as a design doc? Do the broadest claims cover the mechanism rather than just your current implementation, or could a competitor achieve the same effect a different way and walk free? And do the dependent claims descend through the variations you identified, so there is somewhere to retreat if the broad claim is challenged? If you can’t map your whiteboard session onto the claims, say so. A draft you don’t recognize is a draft that needs another pass, and a drafting process that discourages that conversation is telling you something.
The compounding effect
None of this is exotic; it’s process. But it compounds. The company whose engineers give one good disclosure hour per invention ends up, three years later, with a portfolio that reads on its products, survives diligence without discounts, and gives litigation counsel something to work with. The company that treats disclosure as paperwork ends up with the same number of patents and a fraction of the asset value, having paid the same legal fees along the way.
Brandon Furdock is a registered patent attorney and computer engineer (M.S., Northwestern University). Chicago Patent Law runs structured invention-disclosure sessions with engineering teams as part of every application engagement and its fractional IP counsel retainers. General information, not legal advice. Attorney Advertising.