A lead enters.
Here’s exactly what happens next.
Most pages tell you what an AI sales rep can do. This one shows you what it does. Step by step. From the moment a form is submitted to the moment a closer picks up a warm meeting on Tuesday at 2:15.
A lead lands. Within 11 seconds, DialEcho knows more about them than your last rep did after a discovery call.
Form submit captured
Webhook fires in 80ms. Lead drops into the pipeline raw.
Enrichment cascade
Pulls firmographics, tech stack, funding, and intent signals from 14 data sources in parallel.
Behavioral scoring
Cross-references 6 weeks of site activity. Visited pricing 4×? That's a 92.
Routine routing
Matches the lead against your active routines. Picks the highest-confidence match.
It picks a routine. Routines are the rules of the game — written by you, run by DialEcho.
Targeting · script · channel · policy.
routine "warm-pricing-visitor" {
trigger: pricing_visits >= 3 within 7d
channel: voice → sms → email (cascade)
voice {
opener: "Hey {{first}}, this is Echo from DialEcho..."
objection_map: pricing_too_high → tier_calculator
book_against: closer_calendar.warm_inbound
}
policy {
timing: business_hours(prospect_tz)
max_attempts: 3 across 5 days
handoff_if: hot_intent | manager_request
}
}Routines are versioned, A/B-testable, and editable in plain English or YAML. Forking the one that’s winning takes one click. Edits ship live in seconds — no engineer required.
It places the call. Sub-500ms latency. Handles interruptions. Books the meeting before hanging up.
Sub-500ms response
Faster than the average human pause. Prospects can't tell.
Interruption handling
Stops mid-sentence the moment Jamie talks. Resumes from context.
Objection map
47 pre-trained rebuttals per routine. Custom ones learn in days.
Calendar awareness
Knows Sarah's calendar in real time. Won't book over a meeting.
Multi-language
Same agent, 11 languages. Switches on detection.
It choreographs the rest. SMS, email, calendar, Slack — all from the same brain. Then it gets out of the way.
What’s actually running.
Speech stack
Streaming ASR with semantic VAD. Voice synth fine-tuned on 12,000 hours of inbound/outbound sales calls.
Reasoning layer
Multi-step planner runs in parallel to speech. Picks next move (rebut · pivot · book · escalate) every ~2 seconds.
Memory
Per-lead memory persists across channels. The SMS knows what the call said. The email knows the SMS.
Tool layer
First-class integrations: Salesforce, HubSpot, Slack, Google Cal, Twilio, Zoom, Stripe, Snowflake, Notion.
Policy engine
Per-routine and per-account guardrails. TCPA timing, dialing windows, compliance flags, escalation triggers.
Eval & replay
Every call evaluated against 12 dimensions. Failures auto-flagged. Replay any conversation with one click.
The smartest move is knowing when to stop.
Hot intent detected
Buyer says 'I need to talk to someone today.' Slack ping to closer, call patched live.
Pricing > $50k ACV
Anything above your enterprise threshold escalates to a named senior rep.
Legal / contract questions
MSA, DPA, redlines — never the AI's job. Hands to your legal contact.
Frustration signals
Detected sarcasm, raised voice, or 'speak to a manager.' Patches immediately.
Explicit human request
'Can I talk to a real person?' → yes. Always. No friction.
Lie about being AI
If asked 'Are you a real person?' — it tells the truth. Every time. No exceptions.
Call before 9am or after 8pm
TCPA timing, in the prospect's local timezone. Hardcoded.
Talk to anyone who's opted out
DNC list scrubbed before every dial. Opt-outs honored across all channels in 60s.
Make promises it can't keep
Pricing, contracts, timelines — bounded to your approved messaging.
Record without consent
Two-party-consent states get the disclosure. No recording without it.