02 / 04
THE ENGINES

7 Engines.
One Unstoppable System.

Each engine is powerful alone. Together they form a closed loop that learns, adapts, and compounds — getting better every single day it runs.

🎯 Strategy 📞 Dialer 🤖 Voice AI 🔀 Branching 🧠 ML Engine 💬 SMS A/B ♾️ Follow-Up
01
🎯
Engine 01
AI Strategy
Planner

Tell it your goal in plain English. It builds the entire workflow, scripts, timing rules, and playbook — automatically. No configuration. No consultants. Minutes, not weeks.

Goal-to-workflow in seconds — “Book solar consultations” becomes 19 rules, 4 sequences, and a full playbook
Industry-aware templates — solar, insurance, real estate, debt, B2B. Each one tuned differently.
Self-updating playbook — as data comes in, the strategy adjusts. No manual editing required.
Multi-goal campaigns — run speed-to-lead, database reactivation, and nurture simultaneously.
strategy-planner.ts
// Input: plain English goal
goal: "Book solar consultations"

// Output: full campaign config
workflow: {
  speed_to_lead: true,
  call_within_minutes: 5,
  script_variants: 3,
  follow_up_rules: 19,
  nurture_sequences: 4,
  objection_handlers: 12
}

// ✓ Campaign ready. Activating.
02
📞
Engine 02
Autonomous
Dialer

Not just a power dialer — an intelligent calling system that decides who to call, when, from which number, at what speed. Every decision is data-driven.

50–600 calls per minute — scales from a test campaign to full enterprise volume automatically
Phone number rotation — multiple numbers per campaign, quarantined before they hit spam thresholds
Smart pacing engine — auto-adjusts speed based on live answer rates, error rates, and carrier signals
Callback precision — “Call me Tuesday at 2pm” logs the exact time and fires with full conversation context
Time-zone aware — never calls Florida at 6am PST. Legal calling windows enforced automatically.
// LIVE DIALER STATUS
247
CALLS/MIN
16%
ANSWER RATE
8
ACTIVE #s
3
QUARANTINED
✓ Pacing optimal — adjusting to 253 calls/min
✓ Number health: 5 clean, 3 cooling, 0 flagged
✓ 14 callbacks queued — next fires in 4 min
03
🤖
Engine 03
AI Voice
Agents

Powered by Retell AI + Telnyx. Not a robocall — a full, natural conversation that handles objections, books appointments, and leaves voicemails indistinguishable from a human rep.

Natural voice with personality — choose voice, tone, pacing. Sounds like a trained closer, not a bot.
Full context awareness — knows the lead's name, company, timezone, last conversation, disposition history
Voicemail detection — detects VM in real time, leaves a natural message, schedules retry automatically
Live appointment booking — pulls calendar availability, offers slots, confirms in real time
A/B script testing — winning scripts automatically get more call traffic. Losers rotate out.
live-call-transcript
// Outbound call — Lead: Marcus T.

AI: "Hey Marcus, this is Sarah calling from
    SunPath — you filled out our form
    about solar earlier today?"

MARCUS: "Yeah, just looking around..."

AI: "Totally, no pressure. Quick question —
    you're in Aurora right? Average bill?"

MARCUS: "Like $280 a month..."

AI: "Okay so you're a great candidate —
    most of our Aurora customers cut that
    by 70-80%. Can I grab 20 minutes
    Thursday afternoon?"

→ BOOKED: Thu 2:30pm ✓
04
🔀
Engine 04
Workflow
Branching

Real if/then/else logic with 13 condition operators. Not linear drip sequences — actual branching decision trees that route every lead based on exactly what happened in every interaction.

13 condition operators — equals, greater_than, contains, between, exists, starts_with, and more
Retry loops — attempt 3 times, then escalate, then nurture. No manual intervention.
Infinite nurture cycles — leads that don't convert today enter a legal, perpetual follow-up loop
Interest scoring branches — interest level > 7 → hot queue. 4–7 → warm nurture. <4 → long drip.
Call Answered?
Voice agent completes conversation → scores interaction
↓
YES → Interest > 7?
NO → Send SMS nudge → Retry in 4hrs
↓
YES → Book appt → Confirm → Remind
NO → Follow-up call in 24hrs at best answer hour
↓
3 Attempts Failed?
→ Enter monthly nurture loop ♻ — value SMS, channel rotation, patience
05
🧠
Engine 05
Predictive
ML Engine

Five models running simultaneously on your data. Logistic regression trained weekly on your call history. Chi-square significance testing so it only acts on patterns that are statistically real — not noise.

9 input features per lead — time since opt-in, call attempts, engagement score, timezone, prior response, and more
Trains weekly on your specific call data — not industry averages. Your patterns.
Convergence-based training — model only updates when it reaches statistical stability
Chi-square testing — won't change call timing unless the pattern is proven, not lucky
ModelPredictsImpact
ConversionP(lead will convert)Prioritize high-probability leads first
ChurnP(lead is about to be lost)Auto-triggers re-engagement sequence
IntentTimeline + budget signals2× score boost for “ready to buy” leads
TimingBest hour/day per leadCalls happen when they'll actually answer
MessageWhich SMS copy works per segmentProven winners get more sends automatically
06
💬
Engine 06
SMS A/B
Testing

Every text message is an experiment. The UCB1 bandit algorithm continuously shifts traffic toward winners, rewrites losers using AI, and tracks what actually moves leads forward — not just open rates.

UCB1 bandit algorithm — mathematically optimal traffic splitting. Not 50/50 guessing.
Tracks real outcomes — reply rate, positive sentiment, appointments booked, opt-outs
AI rewrites losers — variants with <5% reply rate get automatically rewritten by the AI
Max 4 variants — focused testing, not scattered. One clear winner per context.
Context-aware — different variants for different lead stages, industries, and dispositions
// SMS A/B TEST — ACTIVE
VARIANT A — WINNING 18.4% reply
“Hey [Name], quick question about your solar quote — did you get a chance to look at it?”
72% traffic → this variant
VARIANT B 9.1% reply
“Hi [Name], we have an exclusive solar deal for Aurora homeowners this month...”
28% traffic → testing validity
07
♾️
Engine 07 — The One That Never Stops
Perpetual
Follow-Up

This is what separates Dial Smart from everything else. Every lead — no matter the disposition — gets routed into the right pipeline and followed up with. Tomorrow. Next week. Next month. Until they buy or legally opt out. That’s the only exit.

Adaptive timing gaps — starts at 7 days, grows to 30 days as leads age. Never spammy, always present.
Channel rotation — SMS → call → SMS → call. Each touchpoint different enough to feel fresh.
Value-first messaging — tips, insights, relevant news. Not pitches. Builds trust over time.
Pipeline-aware routing — good calls that didn't book go to hot pipeline with full context. Callbacks fire on time, to the minute.
Legal stop conditions — DNC, “not interested”, unsubscribe. Three ways out. Nothing else removes a lead from follow-up.
// PERPETUAL FOLLOW-UP TIMELINE
DAY 1
Didn't book → “Interested, needs time” → Callback Day 3, 2pm
DAY 3
Callback fires with full context → Still not ready → 7-day nurture
DAY 10
Value SMS → Reply detected → Moved to hot queue immediately
DAY 30+
Monthly check-in. Channel rotation. Patience.
∞ FOREVER
Until they're ready. The AI never forgets.

// Next Step

See How We Stack Up.

Feature-by-feature vs VICIdial, Five9, and GoHighLevel.

View Comparison → Calculate ROI