An AI SDR that books meetings, without the spam, the burnout, or the lawsuit.
A human SDR costs $5-8k/month all-in and ramps for a quarter. This agent does the tedious 80%, building the target list, researching each prospect, writing genuinely personalized outreach, running the follow-ups, classifying every reply, and hands a warm, briefed meeting to a human closer. It never sells and never closes; it books. And it's built so you don't get blacklisted or sued doing it.
1 The outbound motion, end to end
2 Why it's built this way, the rationale
The high-leverage choices below reflect outbound & deliverability best practice BENCHMARK; everyday spacing is standard sequence practice STD. Treat any percentages as directional, validate against your own send data.
🔎 Relevance beats volume
Cold outreach fails on relevance, not effort. The agent researches a real trigger (a raise, a hire, a CRM switch) per prospect, and refuses to send generic spray when it can't find a genuine angle. Quality protects your reputation.
BENCHMARK📬 Deliverability is survival
Verify before sending, skip risky addresses, throttle volume per domain, and auto-pause if bounce/complaint rates spike. One blacklisted domain ends your outbound, the agent guards it like an asset.
BENCHMARK🔁 Follow-up is where deals leak
Most replies come on touch 2-4, yet humans forget to follow up. The agent never does, and the moment a human replies, it stops the sequence and hands off, so nobody gets a robotic bump mid-conversation.
BENCHMARK🎯 Trigger-based timing
Reaching a prospect right after a funding round, a key hire, or a tooling change lands far better than a random cold touch. The agent watches for those signals and times outreach to them.
STD🤝 Book, don't sell
The agent's job ends at a booked, briefed meeting on a human AE's calendar. It never negotiates, prices, or closes, that keeps a human in every deal and avoids the agent over-promising.
STD♻️ Polite persistence, then stop
A capped sequence ending in a graceful breakup, plus reason-based nurture for "not now". No endless bumping, over-contacting is how you become spam.
STD3 The guardrails, what keeps outbound safe & legal
📜 CAN-SPAM & GDPR by default
Every email carries a one-click opt-out and a physical mailing address; lawful basis (B2B legitimate interest / consent) is recorded; unsubscribes are honored instantly and permanently.
📬 Deliverability protection
Email verification before send, risky/catch-all addresses skipped, per-domain send caps & warmup, and automatic pause if bounce or complaint rates cross a threshold.
🚫 Suppression & do-not-contact
Unsubscribes, bounces, complaints, existing customers, open deals, and competitor domains are never emailed. The agent won't poach your own funnel or torch a relationship.
✋ Reply → stop · Negative → human
Any human reply pauses the sequence. Interested → book; objection → a thoughtful reply with an out; anger or a legal threat → escalate to a person, never argued with by a bot.
🧭 No false claims, no tricks
No fake "Re:" reply threads, no fabricated "as discussed", no invented stats or logos, no impersonation. Truthful about the product or it doesn't send.
👤 Human approval & handoff
Drafts are approved by a human before send, and booked meetings arrive with a full brief for the AE. The agent does the volume; people own the relationships.
How it works, from a name on a list to a booked meeting.
Every prospect runs the same pipeline. The agent builds and scores the list, researches each person, writes and sends personalized touches, classifies replies, and books meetings, with a compliance & deliverability gate in front of every single send.
1 The pipeline
2 The compliance & deliverability gate, 11 checks
3 The cadence map, every step, when it fires, and why
| Stage | Step | When | Goal | Basis |
|---|
Each step has a wait-window; the send-ledger + gate guarantee once-per-step and stop-on-reply, and every send + reply is logged to the CRM for a full activity trail the AE inherits.
The outreach.
Every scenario below is a real moment in an outbound motion, mapped to a prospect in the sample database. Pick one to see the exact message the agent drafts, the data it researched, the guardrails it applied, and the decision it made (including when it refuses to send, or books a meeting). The last group shows the same engine re-skinned for other verticals.
The database.
The agent reads from your prospecting / enrichment / CRM stack and owns a few tables of its own for sequencing, suppression and deliverability. This mirrors a real outbound stack so the demo can be pointed at live data and a real sending domain in a later phase.
! Critical data rules baked into the agent
Sample data.
A spread of demo prospects across the outbound funnel, freshly targeted, in-sequence, replied, booked, in nurture and suppressed, plus a few from other verticals. This is the exact data the outreach in this demo was generated from. In production it's a live view of your prospecting tools + CRM.
| Prospect | Title | Company | Vertical | Size | ICP fit | Trigger / hook | Stage | Email status | TZ |
|---|
Per prospect the agent sees: account firmographics · role & seniority · verified email · ICP fit score & reasons · trigger/intent events · sequence enrollment & touch history · replies · meetings · suppression status. All names, companies and events are illustrative.