| Thin-waist architecture can generalise architecture · G2 | Open Designed and exercised against synthetic fixtures. No independent production integrations. | - Architecture specification
- SDK implementation
- Synthetic fixtures
- Research programme Paper 1
| Independent heterogeneous production integrations. | A second or third independent source requires customer-specific core semantics. | Three materially different upstream systems emitting through the same primitives. |
| Rust verification is portable architecture · G4 | Demonstrated Deterministic verifier runs against fixture bundles in a clean environment. | - Rust verifier
- Conformance fixtures
- Deterministic test suite
| An external party reconstructing a bundle in an environment AgEvidence never touched. | The offline verifier requires proprietary hosted state to resolve a bundle. | External reconstruction by a reviewer using only the published package. |
| Python integration can remain thin architecture · G2 | Open Internally demonstrated. No external developer has integrated unassisted. | - SDK primitives
- Documentation
- Internal integrations
| An unassisted external integration with recorded time-to-first-artifact. | Every integration requires AgEvidence engineers to model the customer's domain. | One external developer team, no support channel, measured. |
| Rails implements the recurring operational workflow commercial · G5 | Demonstrated Internal product path demonstrated end to end. Recurring external use and willingness to pay are open. | - Working evidence plane
- Profile maintenance model
- Review and issuance workflow
| A renewal, or a second evaluation cycle paid for by the same customer. | Customers pay once at implementation and see no reason to continue. | One paid deployment carried through a second evaluation cycle. |
| Scientific validity stays separated from integrity science | Proposed Designed into the model. Empirical test of user behaviour is open. | - Scientific lineage dossier
- Status vocabulary
- Statement design
| Observation of how real relying parties interpret a passing verification. | Users routinely treat integrity results as scientific or regulatory approval. | Structured interpretation study with external reviewers. |
| One evidence chain can serve two uses architecture · G3 | Open Architecturally supported, never demonstrated with two real destinations. | - Profile separation design
- Fixture coverage
| Two real downstream uses from one chain. | The second destination requires the source evidence to be re-derived. | One chain, two destinations, no source rework. |
| Funded agtech will pay commercial · G5 | Open Market research and pricing work only. No invoiced instance. | - Pricing dossier
- Buyer interviews
- Market structure analysis
| Paid production conversion. | Developers adopt, and budget owners repeatedly refuse paid evidence operation. | One named startup pays for readiness or deployment. |
| The second deployment is cheaper commercial · G6 | Open No second deployment exists. | - Reusable profile design
- Connector model
| Two comparable deployments with recorded hours. | Customer three still requires substantial bespoke plane code. | Instrument deployment hours from the first engagement onward. |
| Open core strengthens distribution commercial | Supported Argued from precedent in adjacent categories. Untested here. | - Precedent analysis
- Open/commercial boundary specification
| A customer who could self-operate the open layer and still pays for managed operation. | Adopters run the open core themselves and decline the commercial plane indefinitely. | Track self-hosting adopters and their conversion behaviour. |
| Portfolio-assisted heterogeneous validation accelerates proof commercial | Proposed Structurally plausible. Discounted by relatedness until an external integration exists. | - Portfolio laboratory analysis
- Wave 0 design
| At least one arm's-length comparator: a non-portfolio integration run in parallel. | Portfolio integrations take as long as cold external ones, or never generalise. | Run one portfolio and one external integration in parallel and compare hours. |
| The protocol can survive without the SaaS governance | Built Architecturally testable today; stewardship arrangements are not yet executed. | - Open licence map
- Offline verifier
- Published fixtures
| Executed stewardship and contribution governance. | Any part of the core requires the hosted plane to function. | Cold-start the core from published artefacts alone, with no company resources. |
| The company survives a failed next financing governance | Open Depends on execution and on hiring being gated to proofs rather than to the vision. | - Capital plan gated to proof
- Lean operating model
| Observed spend discipline over two or more gates. | Payroll grows ahead of gate completion. | Compare headcount additions against gate completion at month 6 and month 12. |