idea-002 · feasibility deck
Every slide states a condition that would have to be true, then reports where it stands using the candidate file's own claim and its own confidence tag. A deck never upgrades a tag, invents a number, or recommends anything, and there is no ask slide.
The chair: the engineers who have to make an adhesive hold vacuum on fur, on a limb, on an animal that moves.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-002.md — the number is the claim's position in its ## Claims list, the tag is copied from it and never adjusted here. A [no claim] marker means nothing in the file speaks to the condition at all. See docs/deck-spec.md.
What changed since the 2026-08-24 build: every bench condition this deck put on record went through the Verifier and came back unconfirmed with the same disposition — "not checkable by any connector in this repo", because they describe hardware that does not yet exist. That is a structural answer, not an unspent check. The research lane also added a recall analysis of the human product code that speaks directly to the canister-free architecture [claim 37: unconfirmed]. Rebuilt from the file, not patched.
Would have to be true: A battery pump would have to hold roughly -80 mmHg through a superabsorbent dressing, across a fur-compatible drape and a limb-conforming interface, on open or grafted distal-extremity wounds, for the duration of a treatment cycle.
Where it stands: The target pressure and the architecture are stated in the candidate's frontmatter; no claim in the file states a duration, a leak-rate tolerance or a seal specification [no claim], so this deck states none either. The one performance requirement on record is the seal-hold condition this deck put there [claim 28: unconfirmed]. One new datum argues the architecture is the right one: 51% (19 of 37) of FDA recalls for canistered NPWT pumps over 2011-2026 name a canister, dressing-connector or tubing defect as the stated reason [claim 37: unconfirmed] — i.e. half the recorded field failures in the human product code attach to components this design removes.
What would settle it: The human predicate's published operating specification (the PICO family) as a starting envelope; a stated treatment-cycle duration from the veterinary case series the file cites [claim 8: verified], now read at abstract depth [claim 35: unconfirmed], [claim 36: unconfirmed].
If it's false: There is no spec to build or test against — and unusually for this repo, no regulator will supply one either [claim 2: verified].
Would have to be true: NPWT's benefit would have to survive translation into this species, this site, and this canister-free format.
Where it stands: The mechanism itself is not in doubt; the format and site are, and the evidence is now mixed rather than uniformly discouraging. Against: the canine pilot is n=5, on 2cm x 2cm hemithorax wounds rather than distal limb, with control wounds better on every macroscopic and histologic measure named and the authors concluding benefits "remain unknown" [claim 9: unconfirmed]; and at this candidate's exact site, a controlled 10-horse study found limited added value versus calcium alginate [claim 20: unconfirmed], echoed by a 2026 commentary reporting isolated PICO-system NPWT insufficient to stimulate healing [claim 39: unconfirmed]. For: a matched-pairs comparison of 50 dogs found significantly fewer complications (p=0.008) and less sepsis (p=0.016) than foam dressing [claim 34: unconfirmed], and the human systematic review of 29 studies (n=5,614) found the single-use canister-free format itself reduced infection, dehiscence, seroma and necrosis [claim 33: unconfirmed]. Note what the tags mean: a literature result is a ScanResult, which carries no confidence tag by construction, so no literature finding in this repo can ever be verified as a finding — only its bibliographic existence can [claim 8: verified], [claim 10: verified].
What would settle it: A human reads the full texts — design, power, site, endpoints — and then a prospective controlled series in the actual target site with the actual interface. The 50-dog comparison is the one whose appraisal would change the most.
If it's false: The engineering problem is real and solved and still does not produce a product, because the therapy does not outperform a dressing that costs a fraction as much.
Would have to be true: Nothing forces anything — and the build plan would have to be written as if something did.
Where it stands: This is the inversion that defines the candidate. Devices used exclusively in veterinary medicine need no establishment registration and no device listing [claim 2: verified]; the premarket exemption itself rests on a Compliance Policy Guide that no connector here can read [claim 1: unconfirmed]; post-market misbranding and adulteration authority is asserted but unchecked [claim 3: unconfirmed]. The human equivalent would be Class II under product code OMP [claim 4: verified] — which is a useful map of the tests a regulator would have demanded, and now something more concrete than that: the recall history under that same code is readable, and it says half the field failures are in the fluid path [claim 37: unconfirmed]. Sterile packaging and tooling validation to a standard that supports the label is on record and structurally uncheckable here [claim 31: unconfirmed].
What would settle it: The human OMP predicate's test battery as a voluntary specification — with the recall reasons as a prioritised list of what to test hardest; regulatory counsel on what labelling claims are defensible without premarket review.
If it's false: The risk here is not a blocked pathway, it is an unforced one: with no reviewer asking for bench data, the discipline has to be self-imposed, and the failure mode is shipping on a seal that was never characterised.
Would have to be true: An adhesive drape would have to seal against fur, over a conical distal limb, through days of ambulation, without irritating the skin enough to break the seal.
Where it stands: Unbacked [claim 28: unconfirmed], [claim 29: unconfirmed], [claim 30: unconfirmed], and the Verifier's disposition on all three is the same: bench claims about a component that does not yet exist, which no registry, regulatory or literature connector reaches. The published record already describes this failing in exactly this way: a complete seal required adhesive spray, vacuum maintenance "lessened over time", and self-limiting skin irritation in all dogs hampered attainment of negative pressure [claim 9: unconfirmed]. That is still the single most useful sentence in the candidate file for an engineer, and the file flags [claim 30: unconfirmed] as the highest-value of the four bench items for that reason. It is also why the moat claim points where it does — fur-compatible drape chemistry and limb interface, with the pump conceded as commodity [claim 16: unconfirmed].
What would settle it: Seal-retention testing on cadaveric or model limbs through a motion cycle at the target pressure; adhesive-tissue irritation testing on furred skin; leak rate over 72 hours.
If it's false: There is no product. Every other condition on all three decks is downstream of this one holding.
Would have to be true: Seal retention, leak rate, pressure stability under motion, and adhesive irritation would all have to be characterised on models before an animal is enrolled.
Where it stands: No bench claim existed in the file before this deck's write-back [claim 28: unconfirmed], [claim 29: unconfirmed], [claim 30: unconfirmed], and a Verifier pass has now confirmed none of them is reachable by tooling — they are bench work or nothing. The candidate's capital plan goes straight to prospective teaching-hospital case series [claim 18: unconfirmed], which is the right eventual step and the wrong first one: the failure mode the literature reports is mechanical [claim 9: unconfirmed], and mechanical failures are cheaper to find on a model limb than in a case series. The equine series makes the cost of getting this wrong concrete — 26% of 42 horses were discharged before NPWT was considered ideally complete [claim 36: unconfirmed].
What would settle it: A limb-model rig with instrumented pressure logging and a motion cycle. No animal, no owner, no institutional review.
If it's false: A case series would be spent discovering a seal problem that a bench rig would have shown in a week — and the profession would read the result as the therapy failing rather than the drape.
Would have to be true: A single-use sterile consumable would have to be manufacturable at a cost that leaves margin, applied correctly by ordinary practice staff, and survive an animal and a bandage over it.
Where it stands: Tooling and packaging validation is on record and unchecked [claim 31: unconfirmed], as is margin at the proposed price [claim 27: unconfirmed] — and those two are the same question from opposite ends. Application by ordinary technicians rather than by study authors is unbacked [claim 23: unconfirmed], and the interface surviving normal animal behaviour likewise [claim 29: unconfirmed]. The human tolerance proxy is worth carrying into the use-design work: 20% of human NPWT patients ceased therapy prematurely at their own request, citing limitations in daily activities [claim 32: unconfirmed], and a dog cannot report the same complaint any way other than by removing the dressing. Notably absent from the candidate entirely: battery life, disposal, and what happens when a standard bandage is applied over the dressing [no claim].
What would settle it: A contract-manufacturing quote against a defined dressing spec; application testing with practice staff; a use-related risk analysis for the home phase.
If it's false: Unit economics fail at the same point the seal does — in the hands of the person applying it, in a building nobody from the company is standing in.
Would have to be true: The foundational NPWT patents would have to be expired, and the live single-use canister-free family would have to leave room for this design.
Where it stands: Neither half is checked, and the two halves fail differently. The foundational Wake Forest patents returned unconfirmed for a missing credential rather than a missing answer — the check recorded in the claims names PATENTSVIEW_API_KEY as unset [claim 5: unconfirmed], [claim 6: unconfirmed], and the repo's IP connector has since migrated to EPO Open Patent Services, so the credential to supply now is EPO_OPS_KEY/EPO_OPS_SECRET. The half that actually matters is worse: the live canister-free / superabsorbent family and the dressing-interface chemistry are "not checkable as written", because the connector takes a specific patent number and cannot run a landscape search [claim 7: unconfirmed]. And the exposure is concentrated exactly where the moat is [claim 16: unconfirmed] — the drape chemistry and interface are simultaneously the only defensible asset and the most likely infringement surface.
What would settle it: Supply the EPO OPS credentials for the expiry checks; a hand search or counsel for the live family, which no credential fixes because no connector here performs discovery; a Tier 3 legal opinion to settle it.
If it's false: Automatic kill under the Stage 5 gate — and on this candidate it is the only barrier of any kind, since the regulatory one does not exist [claim 2: verified].
Every condition above with nothing verified behind it.
The pattern here is the inverse of the human candidates in this repo: the regulatory questions are the settled ones [claim 2: verified], [claim 4: verified], and everything genuinely unknown is engineering and evidence. That is what a market with no premarket review looks like from this chair — the burden does not disappear, it moves onto you. The eight studies added since the last build changed no tag, and could not have: the connector that found them cannot appraise them.
If only one thing from this chair could be checked: whether a fur-compatible drape can hold seal at the target pressure on a distal limb through days of ambulation [claim 28: unconfirmed].
Everything the candidate claims as its own — the moat [claim 16: unconfirmed], the reason to buy this rather than the human unit already in the building [claim 19: unconfirmed], the technician's ability to apply it [claim 23: unconfirmed] — reduces to this one mechanical property. The published literature reports it failing in precisely this way [claim 9: unconfirmed], and the human recall record says the fluid path is where this therapy breaks in the field generally [claim 37: unconfirmed]. It is answerable on a bench, this quarter, with no animal and no regulator involved, and the Verifier has confirmed there is no shortcut: no connector in this repo will ever answer it.
Naming it is not a recommendation, a gate, or a kill.