V
AgentOS by VitalEdge
Capabilities Catalog — Modular AI Infrastructure
Catalog
V. 2026.04
Private Preview
What We've Built · April 2026

Everything we've built,
laid out
à la carte.

A complete catalog of the AI capabilities we've engineered for independent insurance agencies. Voice AI, live in-ear coaching, call audit, personal assistant layer, training infrastructure, virtual floor, compliance rails — the whole operating system, broken down by module. Pick what you need. Skip what you don't. Every module is independent. Every module is white-label. Every module was engineered, not glued together.

3
AI operators ·
Name & voice are yours
50+
Modular capabilities ·
All engineered independently
7
Core infrastructure
providers integrated
100%
White-label ·
Your brand, your voice
01 · The AI Operators 02 · Voice — Client 03 · Coaching 04 · Call Audit 05 · Training + Virtual Floor 06 · Assistant Layer 07 · Retention 08 · Compliance 09 · Ops Infrastructure 10 · How It's Built
01

The three AI operators.

We've built three specialized AI operators — each with a defined role, a locked character brief, a pinned voice, and strict boundaries on what it will and won't do. The names below are our working names. You pick what they're called in your deployment, what voice they speak in, and how they introduce themselves. One client-facing. Two internal. None of them share scope with each other.

Client-facing voice
Working name
AVA
AI Voice Agent

The only operator that talks to clients. Answers inbound calls. Calls out to new leads. Qualifies, captures consent, warm-transfers to the right agent. Speaks under your agency name in a voice you select.

Built-in capabilities
  • Inbound phone line answering
  • Outbound lead callbacks
  • Qualification + consent capture
  • Warm transfer to live agents
  • Appointment booking
  • Voicemail handling + callbacks
  • DNC + TCPA compliance built in
Agent-facing coach
Working name
Coach Vance
AI Sales Coach

Lives in every agent's earpiece and on their dashboard. Coaches live during real calls. Scores every recorded call. Runs voice-based roleplay with realistic prospect personas on demand. Never speaks to clients — only to your agents.

Built-in capabilities
  • Live in-ear whisper coaching
  • Post-call voice debriefs
  • Voice-based prospect roleplay
  • Call scoring against your rubric
  • Objection-handling drills
  • Pre-call prospect briefs
  • Coaching plays configurable per agency
Personal assistant
Working name
Eden
AI Executive Assistant

Per-agent scoped. Morning briefings. Drafts every follow-up message. Tracks licenses, book health, policy anniversaries. Never talks to clients. Never auto-sends — every draft waits for agent approval.

Built-in capabilities
  • Daily morning briefings
  • Follow-up email/SMS drafting
  • Calendar orchestration
  • License renewal tracking
  • Book health alerts
  • Pre-call client research pull
  • Policy anniversary reminders
Names & voices are yours to configure

The names above are what we call them internally — AVA, Coach Vance, Eden. In your deployment they can be whatever you want. We've seen some agencies use single-name brands; others use the agency owner's first name; others use abstract labels like "The Line" or "Signal." You pick the names, the voices (ElevenLabs voice library, or bring your own cloned voice), and how each one introduces itself to your team or clients.

02

Voice — Client-facing layer.

The voice stack your clients actually interact with. Engineered for sub-2-second answer latency and full-duplex natural conversation. Compliance rails built into every call path.

Inbound voice agent Voice

Answers every inbound call under 2 seconds. Full-duplex conversation. Qualifies the caller, captures TCPA consent, books or warm-transfers. Configurable greeting, persona, product knowledge.

Outbound lead callback Voice

Opted-in leads only. Fires automatically on every new web lead, form submission, or other documented opt-in source — never on cold lists, never on purchased data, never on DNC-flagged numbers. Configurable SLA (typical deployment calls back under 30 seconds). Handles common objections, books appointment, logs consent artifact with every dial.

Warm transfer flow Voice

AVA stays on the line while connecting the client to an available agent. Hands off context, reason for the call, and any qualifying info captured — so the agent opens informed.

Voicemail handling Voice

Detects voicemail vs. live human. Leaves a configurable voicemail for no-answers. Queues automatic callback windows respecting the prospect's timezone.

Dedicated phone line provisioning

Inbound and outbound numbers provisioned per deployment. Local or toll-free. Carrier-grade routing. Full Twilio + Retell integration handled.

Appointment booking integration

AVA books directly into the agent's calendar during the call. Confirms slot verbally. Sends SMS + email confirmation. Respects TCPA call-window rules.

03

Live coaching — in the ear.

What we think is the most important module in the stack. Agents get coached during real calls — not after. Engineered for under-1-second whisper latency so the hint arrives while it still matters.

Live whisper coaching Voice

Agent is on a live call with a prospect. The coach listens in. When the prospect says something that triggers a coaching moment — an objection, a stall, a high-risk signal — a short voice hint plays in the agent's earpiece. Under 10 words. Playbook-configured.

Voice-based prospect roleplay Voice

Agent dials the coach line and picks a persona. The coach plays a realistic prospect — with a back story, objections, stalls, emotional beats. Agent runs the full call. Coach scores and debriefs at the end. Unlimited practice.

Configurable prospect personas Voice

Built-in library of common agency scenarios (confused senior, price shopper, angry DNC, spouse objection, grief case, etc.) — plus the ability to define custom personas based on your agency's most common prospect types.

Pre-call prospect brief

Before an agent dials out, the coach pulls up what we know about this prospect: prior contact, lead source, known answers, things to avoid. 30-second brief read or displayed.

Post-call voice debrief Voice

When the agent hangs up, the coach delivers a 30-second voice debrief: one thing that worked, one thing to fix, one phrase to use on the next call. Specific and actionable.

Agency-configured coaching playbook

Your senior producers' methods encoded into what the coach says and when. Not a generic coach — a coach trained to teach the way your agency already sells.

04

Call audit — 100% of calls, scored.

Every call — inbound, outbound, transfer — transcribed and scored against your rubric. Not sampling. Not random spot-checks. Every single call.

100% transcription pipeline

Full transcription of every call via Deepgram streaming. Speaker-separated. Timestamped. Searchable. Archived with full audio recording.

Configurable scoring rubric Owner

You define what earns a 10. What gets flagged. What automatically routes to the owner. Rubric can be reconfigured anytime and retroactively re-scores.

Owner-flagged call queue Owner

Calls scoring below your threshold route directly to the owner's review queue — with transcript, recording link, score breakdown, and recommended coaching action.

Retention-risk detection

Audit engine flags calls where the agent closed something likely to fall off the book — wrong-fit product, pressure-tactic signals, unanswered medical questions. Policies at retention risk caught at issue.

Objection-handling scoring

Every common objection is tagged when it appears in a call. Agents get scored on how they handle each specific objection — cumulatively visible over time.

Compliance auto-flag

Specific words, phrases, or missing disclosures trigger automatic compliance flags. TCPA violations, CMS/TPMO issues, carrier rule breaks — caught automatically.

05

Training & Virtual Floor.

The environment your agents actually log in to every morning. Training delivery, live sessions, script library, product knowledge, and the virtual floor where agents work side-by-side remotely.

Virtual Floor environment

Spatial audio-visual workspace where remote agents see and hear each other during the workday. Replicates the energy of a physical sales floor. Configurable rooms, huddle spots, manager walk-around.

Live daily training blocks

Scheduled daily live coaching sessions hosted on the virtual floor. Recorded, archived, and available on demand for anyone who missed it. Integrated into each agent's training calendar.

12-week structured training program

Cohort-based training curriculum pre-built. Weekly modules, quizzes, practice calls, certification at the end. Configurable — drop your own curriculum in or use ours as a baseline.

Custom script library

Your agency's scripts encoded into the platform — by product, by carrier, by scenario. Accessible from the agent dashboard. Version-controlled. Updates push to every agent instantly.

Carrier underwriting knowledge base

"Client has [condition] and takes [medication] — which carrier takes them?" Answered instantly. Pre-loaded with major life + health carriers, configurable for your appointed stack.

Medication + condition cross-reference

Every common prescription drug cross-referenced to carrier underwriting grids. Agent can ask mid-call. Coach retrieves the answer while the conversation is still live.

Recorded session library

Every live training, every notable call, every coaching moment — archived with searchable transcript. Agents can search the library for "how to handle the 'I need to think about it' objection" and pull up every example.

Agent certification tracking

Training completion, quiz scores, live roleplay pass rates, carrier certifications (AHIP, CMS, state-specific) all tracked in one dashboard. Owner sees who's ready and who needs more reps.

06

The personal assistant layer.

Every agent gets their own AI executive assistant. Per-agent scoped — sees only their book, their calendar, their pipeline. Handles the administrative overhead that eats into production time.

Morning briefing Voice

7 AM daily briefing delivered as voice or text. Today's appointments, priorities, alerts, open loops. Agent walks into their day knowing exactly what's on the plate.

Follow-up message drafting

Agent says "draft a thank-you to Mrs. Patterson after this morning's call." The assistant drafts the email, SMS, or DocuSign nudge — pulling the specific details from the call. Waits for agent approval before sending.

Calendar orchestration

Schedules follow-ups, blocks focus time, reschedules when things run over. Respects agent preferences (no meetings before 9, no back-to-backs, etc.).

License & renewal tracking

Tracks every state license, E&O policy, AHIP certification. Alerts the agent with enough runway to renew without gap. Pre-drafts renewal emails if needed.

Book health monitoring

Monitors policies for payment failures, lapse risk, policy anniversaries. Surfaces the things that need attention today — not in a report, in the daily briefing.

Cross-sell opportunity surfacing

Sees when a Medicare Supplement client should be offered Dental/Vision/Hearing. When a Final Expense buyer fits for Hospital Indemnity. Queues the opportunity for the agent's next conversation.

07

Retention — the whole game.

Writing is one thing. Keeping the policy past month 12 is another. The retention module is built to keep clients engaged from issue date through renewal and beyond.

12-month retention sequence

Birthday, anniversary, AEP reminder, cross-sell touches, referral asks — all automated from issue date. Under your brand. Configurable cadence.

Lapse prediction engine

Flags policies at risk of lapse from payment signals, unanswered contact attempts, recent pattern changes. Save-call triggered before the carrier flags it.

Charge-back prevention

Detects policies likely to NTO or cancel in the first 12 months based on call signals. Triggers a proactive save-call before commission clawback.

Annual review automation

30-60 days before policy anniversary: automated reminder to the client, pre-call brief to the agent, booking flow. Annual review becomes the default, not the exception.

Referral request automation

30 days after issue, a happy client gets asked — by text, email, or voice. Agent can approve the ask or customize it first. Referrals tracked back to source client.

Persistency scorecard per agent

Which agents write clean, 12-month-sticky business vs. who's churning out policies that fall off. Shape commission tiers accordingly. Spot retention leaders.

08

Compliance — built into the rails.

Compliance isn't a layer bolted on top — it's built into every call path, every SMS send, every contact attempt. Engineered so your agents can't break the rules even if they try.

TCPA-compliant dialer rails

8 AM – 8:30 PM local-time enforcement in the caller's timezone. DNC list scrub before every dial. Consent capture logged with timestamp, IP, and consent language version.

A2P 10DLC registration

SMS campaigns registered per deployment. Opt-in language captured. STOP/REMOVE/DNC handling fully automated. Full audit trail for carrier compliance reviews.

Digital business card protocol

TCPA-compliant delivery — sent during the call within the first 60 seconds after verbal permission, never before. Agent's info under the agency's branded card. Swappable delivery mechanism.

Recording + retention policy

Every call recorded. Retained per state, CMS, and carrier requirements. Encrypted storage. PII redaction available (SSN, bank account) before archival.

CMS/TPMO workflow

Scope of Appointment capture, disclaimer language, TPMO-compliant transfer handoffs. Built-in for Medicare-focused deployments.

Compliance escalation queue Owner

Potential TCPA issues, carrier complaints, DNC violations all escalate to the owner with full transcript and context. Nothing sits in a log nobody checks.

09

Ops infrastructure.

The plumbing. CRM build-out, automation flows, data layer, billing, dashboards — done for you, white-labeled, handed over with full ownership.

Full CRM build-out

Pipelines, calendars, automations, tags, custom fields, SMS + email templates. All branded to your agency. Agents see your company name — not ours.

Automation engine

20+ pre-built workflows: lead intake, audit pipeline, compliance escalation, billing reconciliation, retention sequences. Each configurable per deployment.

Operations data layer

One source of truth for agents, clients, commissions, policies, leaderboards, compliance logs. Your data, your account, exportable anytime.

Command center portal

Agent-facing dashboard at a subdomain of your choice. Fast, secure, SSL-wrapped. Agents see their book, pipeline, calls, coaching queue, commission status — all in one place.

Owner live-floor dashboard Owner

See every agent on one screen. Who's live, who's idle, who's flagged. Real-time status. Drill into any agent's calls, pipeline, or coaching queue.

Leaderboards + gamification

Live floor-visible leaderboards. Apps submitted, premium issued, call streaks, team cohorts. Configurable metrics. Visible to every agent.

Call streak tracker

Daily minimums per agent. Breaks alert the agent, their manager, and the owner. No more silent no-shows.

Commission reconciliation

Every carrier statement parsed. Every charge-back caught. Every override verified. Agents see their earned commission in real time. No black box.

Billing + credit-pack engine

Platform subscription billing automated. Usage-based credit packs for call minutes, SMS, AI voice minutes. Stripe-backed, reconciled automatically.

White-label domain + branding

Your domain, your logo, your color palette, your typography. SSL certificates handled. The entire agent experience renders under your brand with no VitalEdge surface area visible.

The Whole Point

Take all of it. Take some of it.
Take one module. Doesn't matter.

Every capability listed above is engineered to work independently. You want just the voice layer because you already have a good CRM? Fine. You want only the coaching and audit modules? Fine. You want the whole operating system? That's the default deployment. Every module slots in with what you already run — or replaces what you don't want to maintain anymore.

10

How it's actually built.

This is the section most AI-for-agencies pitches skip. We're going to tell you exactly what sits underneath, because the difference between real infrastructure and a slapped-together wrapper over ChatGPT shows up the moment you try to deploy to 50 agents. This stack is engineered — edge-deployed, modular, version-controlled, compliance-first.

This is solid hardware and real engineering.
Not a demo, not a slap-together.

Every capability above runs on dedicated infrastructure — no shared multi-tenant bottlenecks, no unversioned prompts, no hardcoded API keys rotated by hand. Every AI operator runs on its own edge-deployed worker. Every voice ID is pinned. Every character brief is version-controlled. Every customer gets their own environment variables, their own billing boundary, their own data isolation.

We chose each piece of infrastructure for a reason. Voice latency matters for live coaching, so voice streaming runs on Retell. Transcription accuracy matters for audit scoring, so every call goes through Deepgram. Voice quality matters for credibility, so TTS runs on ElevenLabs with locked voice IDs. Reasoning quality matters for coaching, so the intelligence layer is Claude Sonnet 4. Compliance telephony matters for TCPA, so call routing runs on Twilio. Edge latency matters for everything, so the orchestration runs on Cloudflare Workers — every request hits a POP within 50ms of the user.

Every one of those choices was made because the cheaper option failed under real load. The result is an AI system that actually holds up when you hand it to an agency — not a demo that dies at deployment.

Edge orchestration
Cloudflare Workers
Every AI operator runs as a dedicated edge worker. Sub-50ms latency to any POP worldwide. Zero cold-start.
Voice infrastructure
Retell · Twilio
Retell for AI voice streaming + real-time transcription. Twilio for compliance telephony + SMS routing.
Transcription & audio
Deepgram
Streaming speech-to-text on every call. Speaker-separated. Used for live coaching triggers + full call audit pipeline.
Voice synthesis
ElevenLabs Turbo v2.5
Low-latency voice synthesis for all three AI operators. Voice IDs pinned per character. Custom voice cloning available at enterprise tier.
Reasoning & language
Claude Sonnet 4
Every scored call, every coaching response, every drafted message routes through versioned character prompts running on Anthropic's Claude.
Workflow automation
n8n
Self-hosted workflow engine orchestrating the long-running processes — onboarding, lifecycle events, lead intake, retention sequences.
CRM surface
GoHighLevel
Agent-facing CRM layer. White-labeled per deployment. Custom pipelines, automations, branded communications.
Data layer
Airtable
Operations data layer — agents, clients, commissions, audit logs. Exportable anytime. Connected through typed API procedures.
Billing
Stripe
Subscription billing + usage-based credit packs. Webhook-signed lifecycle events. Full audit trail per customer.
10b

Engineering principles.

Because “AI-powered” can mean real engineering or a hack with a pretty frontend. These are the principles we hold the product to.

Principle 01

Character briefs are versioned, not vibes-based

Every AI operator runs on a locked, version-controlled character brief. No prompt drift. No "vibes" deployments. Your compliance team reviews and signs off on briefs before go-live. You own the configuration.

Principle 02

Voice IDs are pinned per character

AVA's voice is the same voice on every call, every day, across every deployment. Coach Vance sounds like Coach Vance — not a random TTS default. Brand consistency is built into the rails.

Principle 03

Scope boundaries are enforced in the stack, not in the prompt

The client-facing operator cannot accidentally become internal. The internal operators cannot accidentally leak client-facing. Boundaries enforced at the API layer — not just in the prompt.

Principle 04

Every deployment gets its own environment boundary

Your deployment has its own Cloudflare environment, its own API keys, its own data isolation, its own billing. Nothing shared. Nothing bled. Your agency is not running on shared infrastructure with strangers.

Principle 05

Compliance rails, not compliance features

TCPA timezone enforcement, DNC scrubbing, consent capture, A2P registration — built into the dialer path, not toggled on in settings. Your agents cannot break the rules even if they try.

Principle 06

Every module is independently wireable

The voice module doesn't need the coaching module. The coaching module doesn't need the audit module. The audit module doesn't need the CRM. Take what you need. Everything still works.

If any of this resonates

We'd rather talk about what you'd deploy
than sell you a pitch.

If you run an agency, an IMO, or an FMO and one or two modules on this catalog caught your eye — let's talk. One short call. You tell us what you'd actually use, what you wouldn't, and what you'd want customized. We'll scope it. If it's a fit we build; if it's not, we'll say so.

Co-Founder
Jason Edlin
CEO · VitalEdge LLC
jason@vitaledge.ai
+1 509 580 0308
Co-Founder
Debbie Shinpaugh
COO · VitalEdge LLC
debbie@vitaledge.ai
+1 509 480 9162