DEEPSENSI — ONE-PAGE FACT SHEET (canonical numbers; use exactly as written)
============================================================================
DeepSensi PBC · Dover, DE, USA · press@deepsensi.com · www.deepsensi.com

WHAT IT IS
  The Cognitive Medical OS — a verified clinical-intelligence system with a
  quantitative, fault-tree-derived bound on AI hallucination.

SAFETY (analytically derived)
  Worst-case bound ......... 3.23 x 10^-6 undetected hallucinations / assertion
                             (about 1 in 309,600)
  Nominal bound ............ 1.69 x 10^-7 / assertion (common-cause adjusted)
  Method ................... Fault Tree Analysis, IEC 61025; 23 independent
                             verification barriers in three tiers
  SIL context .............. meets the numerical demand-mode SIL-4 target with
                             ~31x margin (numerical-target equivalence only)

ACCURACY (empirical; development-cohort result; confirmatory study designated)
  Validation set ........... 301 published NEJM Clinicopathological Conference
                             cases (2014-2023)
  Top-1 .................... 86.0% (95% CI 82.1-89.9)
  Top-3 .................... 93.7% (95% CI 91.0-96.4)
  Missed-critical rate ..... 4.6% -> 0.0%, at a cost of +2.3 additional seconds
  Median deliberation ...... 14.3 s

SPEED
  Full multi-specialist deliberation: 12 seconds to minutes for complex cases vs. days to convene an equivalent human panel by hand.

THE STANDARD
  The DeepSensi Standard (DSS) / Gomola Framework: first quantitative safety-
  certification standard for clinical AI. Five auditable pillars, four levels.
  Open and royalty-free (attribution required). Reference implementation
  is proprietary.

COMPANY & COMPLIANCE
  DeepSensi PBC, a Delaware public benefit corporation.
  8 The Green STE A, Dover, DE 19901, USA.
  Founder & Chief Architect: Tomasz Jan Gomoła (ORCID 0009-0001-5222-6154).
  Zero PII by architecture · HIPAA · GDPR · EU AI Act (Articles 9-17) ·
  FDA Q-Submission engaged · HL7 · FHIR · DICOM.
  IP: 12 patent applications filed, 24 further prepared.
  Deployment: cloud or fully offline / air-gapped, from a $500 edge node up.

PAPERS
  Nine technical papers (WP-001..007, DSS-001, NEJM AI Perspective).
  Preprints headed to medRxiv; manuscripts in submission. Full texts:
  www.deepsensi.com/papers

DOCTRINE
  "We publish the proof, the standard, and the audit. We do not publish the
  blueprint."
============================================================================
