Overview
skyVoice is a human-like AI that manages inbound and outbound calls instantly, with verticalized, PHI-aware IVR menus, intelligent routing, lead capture, and CRM integration. Every call is recorded, transcribed, summarized, graded, and written back to the CRM, turning the phone into a measurable channel alongside web and email.
What It Does
Answers and places calls for intake and screening, insurance verification, appointment reminders, referrals, billing questions, and basic triage. Agents use configurable intake questionnaires, per-department voices, and reusable skills that can be shared across queues.
Touch-tone (DTMF) menus and AI virtual receptionists detect why someone is calling and route them to the right department or straight to a preferred representative at the greeting. The receptionist recognizes departments by name or paraphrase and skips inactive queues. Each queue has its own business hours (including overnight schedules) and after-hours behavior, and on a handoff the receiving agent gets the caller's recent words and recorded consent so it does not ask again.
Admins or reps can set out of office with a return date, message, and covering colleague; it ends automatically at the return time. When a caller asks for someone who is away, the agent explains and offers voicemail or the covering rep, never transferring silently. A per-rep toggle controls whether the AI receptionist can route to them directly.
Creates leads from intake calls and syncs them to skyAdvisor, Salesforce, HubSpot, Dazos, and the Alleva EHR, with lead pipelines, rep assignment, and CSV import. Phone numbers are validated, contact-less entries blocked, duplicates matched in any phone format, email addresses captured from intake answers, and failed syncs are retried and alerted. Where caller ID is enabled, returning callers are recognized by phone number and asked to confirm their name; callback requests become callback leads.
AI outbound campaigns with uploaded contact lists, scheduling, retry logic, and pre-loaded caller context so follow-ups pick up where the last conversation left off. Answering machines get a recorded voicemail, and every outcome links back to its call record and can post results to your systems.
Warm transfers to staff with full context, rolling over to an announced, whisper-screened backup number and then to voicemail rather than a dead line. Multi-party conferences and after-hours voicemail and overflow are supported. Crisis situations are routed to the 988 Suicide and Crisis Lifeline and medical emergencies to 911.
A shared, searchable voicemail inbox with recordings and transcripts, including calls whose transfer or rep patch did not connect. Voicemails can be forwarded to transfer ownership and queues can email new messages to a list. When a rep does not answer a transfer, the voicemail is filed under that rep and emailed to them.
Pulls and validates insurance eligibility, patient records, and appointment data from the Skypoint platform in real time to resolve calls faster.
Handles intent detection, data lookups, and workflows without a human operator. Unclear answers are asked again, and spoken emails and phone numbers are read back and validated before reaching your CRM.
Voice Quality and Tuning
How an agent sounds is configurable per instance and per agent.
| Setting | What it controls |
|---|---|
| Conversation mode | Natural speech-to-speech for the most human turn-taking, or a structured pipeline for tool-heavy, question-by-question capture. |
| Voice and language | Text-to-speech provider and voice (including custom ElevenLabs voices) and the speech-to-text language. |
| Interrupt sensitivity | How readily the agent yields, so filler words and pauses do not cut callers off. |
| Pronunciations | A dictionary for names, medications, and other hard-to-pronounce terms. |
| Greeting and messages | Greeting, consent script, and after-hours and voicemail prompts, editable per queue. |
When something goes wrong, the agent behaves honestly: it apologizes instead of leaving dead air, speaks numbers digit by digit, offers reassurance during long silences, and announces transfers before connecting. It waits for the caller's answer before saving a lead or ending the call, and emergency instructions and callback numbers are always spoken in full.
Call Quality and Prompt Optimization
Every call is scored against configurable rubrics, including an intake funnel rubric. Supervisors can override grades, and auto-tag rules re-apply so reporting stays consistent.
Scores an agent's prompt and intake questions across quality dimensions and suggests a rewrite as a diff. Nothing is applied without human review.
Findings are backed by real call outcomes and transcript excerpts showing where callers hesitated or had to be asked twice.
Prompts and questionnaires are versioned and each call records its version, so you can tell whether a change helped and roll back if not.
Reporting and Data Delivery
Dashboards for volume, outcomes, handle time, department, intent mix, and lead conversion, in each customer's own timezone, with breakdowns by community and routing outcome.
Dynamic number insertion and forwarded tracking numbers tie calls back to their marketing source and location, shown in the call list and analytics.
Build a call, lead, or campaign export once and have it downloaded or emailed on a schedule.
Per-call cost tracking by model and instance, with an editable rate table so finance sees spend per call.
Transferred calls stay attributed to their rep, and the Calls page filters by rep and by answered or missed transfer, with completed warm transfers badged. Answer-rate reporting shows who was dialed and who picked up.
Recordings and transcripts cover both the AI portion and the human conversation after a transfer, with secure download and shareable links. The call detail shows any voicemail left on the call with its audio, transcript, and link.
Compliance and Governance
Built for regulated care settings, with controls enforced on the server rather than left to the AI.
Recording is enabled per queue and gated on spoken consent; declines and revocations are honored immediately. Only consent the caller actually gave carries across a receptionist or IVR handoff, so callers are asked again where needed.
HIPAA-aligned handling with signed BAAs, 42 CFR Part 2 consent, PII masking, field-level encryption, audit logging, and media purged on your retention schedule.
Do-not-call and opt-out suppression, quiet hours with per-state rules, and the consent basis recorded on every outbound attempt.
Crisis language is handled deterministically on every path with a referral to the 988 Lifeline, and do-not-hospitalize and goals-of-care escalations fail closed to a human.
Role-based access down to navigation and voicemails, scoped per instance, and revocable for one instance without deleting the user.
Shared sign-on and an app switcher let staff move between skyVoice and other Skypoint apps without signing in again, with release notes and the user guide available in the app.
Why It's Needed in the Market
Front offices in provider groups, health systems, and senior living communities face high call volume, staffing shortages, inconsistent caller experiences, missed after-hours calls and leads, and no attribution for phone calls.
Generic call bots don't understand healthcare context (MRNs, payer rules, care plans). skyVoice delivers verticalized, PHI-aware voice automation, keeping humans in the loop for complex cases.
Key Metrics It Moves
Instant answers and no hold queues reduce wait times and dropped calls.
Real-time access to patient, eligibility, and appointment data resolves more issues on the first call.
AI handles intake, screening, routing, and follow-up so staff focus on cases that need a human.
Immediate answers and seamless handoffs improve CSAT and NPS.
Every intake call becomes a tracked lead with rep assignment and follow-up.
Closed-loop attribution shows which channels drive calls and true campaign ROI.
Failed transfers, after-hours calls, and unanswered patches land in a tracked voicemail inbox with automatic notification.
Graded calls, versioned prompts, and evidence-backed evaluation make agent quality measurable.