How does a health plan actually pay for this?
The model is a vendor contract — the plan pays a per-member-per-month (PMPM) fee, similar to how Papa and Pyx Health are contracted across 70–100+ MA plans respectively. We are a supplemental/value-added benefit, not a billed medical service. Medicare Advantage plans can include us under SSBCI (Special Supplemental Benefits for the Chronically Ill) or standard supplemental benefits. Medicaid MCOs contract us as an HCBS-adjacent support. PACE programs can bundle us under their capitation. We do not bill CPT codes ourselves — a clinician-billed RTM/CCM pathway is a Phase 2 option for provider-group partners.
Is this HIPAA compliant?
We say "HIPAA-eligible" deliberately. The AWS services underlying Companion Care — Lambda, DynamoDB, Polly, Bedrock — are covered under Amazon's HIPAA BAA program. Full HIPAA compliance is a program, not a product feature: it requires signed BAAs, staff training, and policies on both sides. A signed AWS BAA must be executed before we store real PHI — we have not done this yet for production. On our Enterprise and health-plan contracts we execute a BAA with your organization. We will not claim "HIPAA certified" because no such certification exists.
What does the CY2025 SSBCI evidence requirement mean for us?
Since CY2025, CMS can deny an MAO's bid unless the plan demonstrates with an evidence bibliography that an SSBCI benefit is "reasonably expected to improve or maintain the health or overall function" of chronically-ill enrollees. We build the artifact for that bibliography: validated screening (UCLA-3, PHQ-2/9) with pre/post deltas, care-gap closure data, engagement metrics, and escalation outcomes. This is not peer-reviewed RCT evidence — we do not claim that. It is impact-survey and engagement evidence, the same category of evidence ElliQ, Papa, and Pyx use in their plan contracts today.
Is this a medical device? Does it require FDA clearance?
No. We operate in the FDA "general wellness" lane per the agency's Jan 6, 2026 guidance on low-risk general wellness products. We do not diagnose, treat, or name conditions as findings. We screen, track, engage, and connect members to qualified humans when a screen is elevated. This design choice (the companion always routes to a human for clinical interpretation) keeps us wellness-exempt and simultaneously satisfies the payer's crisis-escalation requirement. A general-wellness posture memo is available for your compliance team.
Do you have clinical results yet?
No — we are in pilot stage with no enrolled members. We will not fabricate outcomes or claim results we have not measured. The evidence we cite for sector ROI ($6.7B isolation cost, $1,608/beneficiary) comes from AARP Public Policy Institute / Stanford analysis, publicly reported. The engagement benchmarks we reference (ElliQ 95%/80% loneliness-reduction, Pyx 2.5:1 ROI) are those vendors' own program impact reports — not our results, and not peer-reviewed RCTs. We cite them as market precedent for the category, labeled as such. Our first pilot will generate the first real data point for this program.
What screening instruments do you use and why those?
We use UCLA-3 (the 3-item Loneliness Scale, Hughes et al. 2004), PHQ-2 / PHQ-9 (Kroenke, Spitzer, Williams), and GAD-7 (Spitzer et al. 2006). All three are public domain and freely usable in practice without license fees — important for a scalable PMPM model. UCLA-3 is the instrument Pyx Health uses at onboarding and is the recognized standard for loneliness screening in health plan contexts. PHQ-9 is the behavioral health screen CMS recognizes for Stars/HEDIS quality measures. GAD-7 pairs with PHQ-9 in the quarterly outcome report.
What is your crisis protocol?
Any conversation containing risk language, or a PHQ-9 with item 9 (self-harm thoughts) scored above zero, triggers immediate escalation. The companion delivers 988 Suicide & Crisis Lifeline information (call or text 988) to the member and raises a CRISIS-severity alert to the caregiver/clinical dashboard in real time. Every escalation is audit-logged. A human reviews every crisis-flagged conversation. This is a hard design requirement, not an optional feature.