Assumptions & Limitations
What the architecture assumes, what has been validated, what has not yet been validated, planned milestones, known limitations, and future work. Disciplined engineering judgment, not overconfidence.
Why this page exists
Mature diligence libraries explicitly state what they assume, what they have proven, and what they have not yet proven. Documents like this often increase investor confidence because they show disciplined engineering judgment rather than overconfidence. This page is an honest accounting of where the platform stands.
What the architecture assumes
Enterprises have idle or underutilized GPU capacity they are willing to contribute to the mesh.
WebGPU will continue to mature as a stable browser standard with broad cross-browser support.
Browser sandboxes will remain reasonably secure against escape attempts — they are the outer containment boundary.
Enterprises will pay for distributed inference at the proposed base rate, calibrated against real utilization.
The mesh scheduler can scale to hundreds of nodes without unacceptable latency or dispatch degradation.
Fiat settlement rails (ACH, SWIFT, SEPA, FedWire, RTP) remain available and reliable for enterprise use.
The regulatory environment remains favorable to a non-speculative Distributed Intelligence Unit model.
Node operators act honestly in the majority — the verification architecture is designed to catch the minority who do not.
What has been validated
WebGPU inference execution works in the browser across supported runtimes.
Stripe billing integration is live and processing real enterprise subscriptions.
Entity schemas for compliance, audit logging, and consent records are implemented and deployed.
NIST 800-63B identity assurance tiers are implemented in the DeveloperIdentity system.
GDPR controls — consent records, data retention, right-to-delete, right-to-export — are implemented as entity schemas.
Burn-and-mint settlement flow is documented with a full comparison against transfer-based models.
Five-layer verification architecture is documented with the proof chain specification.
Cross-system identity sync is validated across DOSFI, MeshInfer, and MNE platforms.
Permission manifest system is implemented with declared data access, routing, and privacy constraints.
The end-to-end transaction flow is documented from compute contribution to fiat settlement.
What has not yet been validated
SOC 2 Type II certification — requires an independent audit that has not yet been completed.
HIPAA certification — the architecture is designed to support HIPAA workloads via privacy tiers, but is not certified.
Scheduler performance at 500+ enterprise scale — has not been stress-tested at that volume.
Multi-party treasury governance in production — currently founding-team control; full multi-party transition is planned.
Base compute rate calibration against real utilization data — the current rate is a starting estimate.
Challenge-response probing at scale — the architecture is designed but not stress-tested under adversarial load.
Browser sandbox escape resistance under dedicated adversarial testing — containment is defense-in-depth, not formally verified.
Independent third-party security audit of the full stack — planned, not completed.
Planned milestones
SOC 2 Type II audit
PlannedHIPAA workload certification path
PlannedMulti-party treasury governance transition
PlannedScheduler stress testing at 500+ node scale
PlannedBase compute rate calibration against real utilization
PlannedIndependent third-party security audit
PlannedFormal verification of settlement ledger invariants
Future workHardware security module (HSM) integration for key custody
Future workZero-knowledge proof integration for privacy-preserving verification
Future workKnown limitations
The closed-core model means external security review of the scheduler, router, and readiness engine is limited to the open surface — SDKs, documentation, and APIs. Internal logic is not disclosed.
The browser sandbox is the outer containment boundary. A browser-level escape would compromise AI runtime containment — but would also compromise every browser-based application, not just this platform.
Current treasury governance is founding-team control, not fully multi-party. The transition to multi-party control with HSM-backed key custody is planned but not yet complete.
The base compute rate is an estimate derived from four factors, not a calibrated rate based on sustained real-world utilization data.
Challenge-response probing is designed and documented but has not been stress-tested at scale under dedicated adversarial conditions.
The architecture is designed to support HIPAA and SOC 2 requirements — it is not certified. Compliance-aware architecture is not the same as certified compliance.
Node operator honesty is assumed in the majority. The verification architecture is designed to detect and exclude dishonest nodes, but cannot prevent a determined adversary from attempting to join.
Future work
Formal verification of settlement ledger conservation invariants.
Hardware security module (HSM) integration for treasury key custody and multi-party signing.
Zero-knowledge proof integration for privacy-preserving compute verification.
Expanded multi-party governance with board-level approval thresholds and emergency pause controls.
Independent third-party security audits on a recurring cadence.
Adversarial red-team testing of the challenge-response probing layer.
Calibration of the base compute rate against sustained real-world utilization data.
Scheduler stress testing at 500+ node and 100+ enterprise scale.
DOSFI Sovereignty — Closed Core / Open Surface
DOSFI Core is sovereign and closed. All educational content on this page is generated exclusively from open-surface materials — SDKs, ABIs, documentation, tutorials, examples, and conceptual explanations. Internal details of the DOSFI Runtime, Scheduler, Router, Readiness Engine, Privacy Model, Power/Thermal Model, Vital internals, DIU internals, MNE internals, and Mesh-Native App runtime internals are not revealed, described, or inferred.
No proprietary algorithms, system code, or architectural diagrams capable of enabling replication of the DOSFI Core are published. When internal logic is referenced, only high-level conceptual explanations are provided. DOSFI's sovereignty, mission-lock, and non-replicable architecture are preserved at all times.