IP and Field-of-Use in a Planning License
IP field of use in a planning software license for OEMs: who owns trained weights and derivative IP, exclusivity, re-training bars, term and exit, negotiated not assumed.
Key takeaways
When an implant manufacturer licenses a vendor-neutral AI planning engine, the deal is only as safe as its contract. Four questions decide whether the arrangement protects you: who owns the trained weights and any derivative IP, field-of-use and exclusivity, re-training bars, and term and exit. None of these should be assumed from a template or a handshake. They should be written down, because the default answer to each is often the wrong one for at least one party. Salnus works with manufacturers on a manufacturer-pays, co-development or white-label basis: Salnus provides the implant-agnostic, CT-based planning software (segmentation, CPAK phenotyping, alignment options and sizing support), you keep your implant relationship and channel, and IP is contract-defined rather than inherited from a default. Salnus is currently Research Use Only (RUO), not a cleared device, so any engagement is a pilot, co-development or licensing collaboration. This post walks through the clauses that make or break a planning license, from a manufacturer's point of view.
Why these are negotiated, not assumed
A planning-software license is not a stock software purchase. The asset is partly the code, partly the trained model, and partly the clinical validation and workflow around it. Because those pieces have different origins, "who owns what" is genuinely ambiguous until a contract resolves it. A manufacturer that treats the standard SaaS terms as good enough can discover, years later, that it has no exclusivity in its own segment, no control over where the underlying model goes next, and no clean way to exit. The fix is not exotic. It is to name each of the four issues below explicitly and negotiate them, rather than let a default fill the gap. For the wider build-versus-license decision that sits above this, see our note on build versus buy for implant makers.
1. Who owns the trained weights and derivative IP
This is the clause manufacturers most often under-specify. In a planning collaboration there are usually three layers of IP: the core engine the vendor brings, any derivative work created during the engagement, and the data and annotations that feed validation. A clean contract says, for each layer, who owns it, who can use it, and for what.
A common and honest structure is that the specialist keeps the core planning engine and its underlying model as background IP, while the manufacturer gets a defined license to use it, plus ownership or a license to the parts specific to its catalog and workflow. The point is not that one arrangement is universally right. It is that the trained weights and any co-developed derivatives are valuable enough that ownership should be stated, not left to the reader. Salnus's stance is that the core engine and derived model weights remain Salnus IP, the manufacturer receives a defined field-of-use license, and catalog-specific work is handled explicitly in the contract. For why that neutrality is the foundation, see implant-agnostic surgical planning.
2. Field-of-use and exclusivity
A field-of-use clause defines the boundaries of your license: which anatomy, which procedure, which geography, and which channel. Exclusivity defines whether anyone else can operate inside those same boundaries. These two clauses are where a manufacturer either protects a real competitive position or quietly gives one away.
The tension is structural. A vendor-neutral engine is valuable precisely because it can serve many catalogs, so a specialist has a strong reason to keep the engine broadly available. A manufacturer, meanwhile, may want assurance that a direct competitor cannot license the identical configured planning experience in the identical segment. The resolution is almost always a scoped exclusivity: exclusive or semi-exclusive within a defined field-of-use, non-exclusive elsewhere. That lets the manufacturer secure its lane without asking the vendor to abandon the neutrality that makes the software useful. This is the same neutrality theme covered in vendor-neutral versus vendor-locked planning, applied to contract language rather than architecture.
3. Re-training bars
Because the licensed asset is a trained model, one clause has no equivalent in ordinary software: what the vendor may and may not do with the model afterward. A re-training bar limits how the underlying model can be re-trained or re-purposed in ways that would erode the manufacturer's position. Without it, a manufacturer could fund a pilot, help refine a workflow, and then watch the improved capability be offered to a competitor.
A reasonable bar does not freeze the engine forever, which would be bad for everyone, including patients. It draws a line: the vendor may keep improving the general engine, but may not train a competing, differentiated planning capability specifically for a rival inside the manufacturer's protected field-of-use, and may not fold catalog-specific derivatives into a competitor's offering. Getting this clause right is what turns a re-training bar from a blunt restriction into a durable protection that both sides can live with. It is closely tied to how the commercial model is structured, which we cover in the manufacturer-pays planning model.
4. Term and exit
The last question is the one people postpone: what happens when it ends. Term sets how long the license runs and how it renews. Exit sets what each party keeps, returns or continues when the relationship stops, whether by expiry, breach, or acquisition of either company. A planning deal without a defined exit is a deal that becomes a hostage negotiation later.
Sensible exit terms cover a few concrete things: continuity of service for plans already in a clinical workflow, return or deletion of data, survival of the field-of-use and re-training protections beyond termination, and clarity on whether the manufacturer can transition off the engine without stranding its customers. For a manufacturer, the practical test is simple. If the vendor disappeared or was acquired by a competitor tomorrow, does the contract say what happens? If not, the exit clause is not finished.
Where Salnus fits
Salnus is positioned as the license and co-development option, not a robot vendor and not a competitor for your implant relationship. The commercial frame is manufacturer-pays: Salnus provides the vendor-neutral, CT-based planning software, you keep your implant relationship and channel, and the four clauses above are treated as the substance of the deal, not boilerplate. Salnus's default stance is contract-defined IP with the core engine and model weights remaining Salnus IP, a scoped field-of-use license to the manufacturer, an explicit re-training bar within that field, and stated term and exit terms. For the white-label version of this arrangement, see white-label AI knee planning for implant companies.
Two honest boundaries. Salnus is currently Research Use Only (RUO), not a cleared device, so an engagement is a pilot, co-development or licensing collaboration, with no clearance, outcome or superiority claim implied. And the contract questions above are meant to be negotiated openly; a fair planning license is one where both the manufacturer and the specialist can point to the clause that protects them.
Bottom line
An OEM planning license is safe or unsafe depending on four clauses: ownership of the trained weights and derivative IP, field-of-use and exclusivity, re-training bars, and term and exit. Each has a default answer that usually favors one side, so each should be named and negotiated rather than assumed. Salnus works with manufacturers on a manufacturer-pays, contract-defined-IP basis, currently Research Use Only, and treats these terms as the core of the deal. To see the whole picture, review which manufacturers offer AI knee planning and how CT 3D preoperative planning works. To explore a pilot, co-development or licensing engagement, see what Salnus offers manufacturers.
Reviewed by the Salnus biomedical engineering team.