ai-cold-outreach

What it does

You are described as an expert in AI-powered cold outreach systems, helping users build, optimize, and scale personalized cold email campaigns that generate pipeline. You cover the full stack from signal detection and enrichment through personalization, sequencing, sending infrastructure, and AI-generated follow-ups, with a bias toward actionable guidance grounded in current data.

How it works

The skill presents a six-stage pipeline for cold outreach:

1. Signal DETECTION: use signals from sources like Clay, Bombora, G2; layer signals for higher intent; Clay orchestrates signal routing.
2. ENRICHMENT: perform waterfall enrichment across multiple providers (Apollo, Hunter, ZoomInfo, RocketReach, Dropcontact, etc.) to obtain a verified work email, then capture title, company size, revenue, recent news, tech stack, LinkedIn URL, hiring signals, and social/AI data. Emails must be verified via a verification service before sending, to keep bounce rate under 2%.
3. AI PERSONALIZATION: use enriched data to generate a research summary, personalization angle, and a custom first line plus a pain hypothesis; merge these into an email template via variables. Use AI models (Claude/GPT family) selected for specific tasks (e.g., first lines, synthesis) and follow recommended first-line frameworks (Observation+Implication, Compliment+Bridge, Trigger+Question, Mutual Connection, Timeline Narrative). Include an AI enrichment step using Clay to produce a 1-sentence observation.
4. SEQUENCING: define a multi-step campaign with days and emails, include conditional branching (opens without reply, link clicks, no opens, positive reply, objections) and A/B testing on subject lines, first lines, CTAs, length, and send time.
5. SENDING INFRASTRUCTURE: detail domain/mailbox architecture with secondary domains and rotation, calculate domains needed from daily volume, provide per-mailbox sending limits, and specify a domain warmup protocol by week. Include an authentication checklist (SPF, DKIM, DMARC, tracking domain, List-Unsubscribe, MX, PTR). Mention DMARC rollout phases and the deliverability benefits of authenticated senders. Compare Instantly and Smartlead on feature sets, pricing, and capabilities, noting native Clay integration and multi-channel focus.
6. CLAY integration and cost: describes how to connect Clay for enrichment, and cost per lead for Sonnet-tier models.

When to use it

Use when the user is building an AI-powered outreach system or mentions cold email, cold outreach, outreach automation, or related terms (Instantly, Smartlead, Clay, email sequences, deliverability, personalization at scale, reply rate, outreach stack). Do not use for technical implementation, code review, or software architecture.

What it can touch

Tools allowed: claude-code, copilot, cursor. The workflow includes interactions with Clay, Apollo, ZoomInfo, Hunter, RocketReach, Clearbit, Bombora, LinkedIn Navigator, and various email verification services, as well as email sending platforms (Instantly, Smartlead) and AI models (Claude, GPT-4o, etc.).

Caveats

No explicit license statements beyond NOASSERTION. The plan emphasizes verified emails and DMARC policy progression, cautioning against high bounce rates and domain burns. It advises conditional branching in sequences and A/B testing with minimum sample sizes, and it warns not to jump to p=reject before inventorying legitimate senders.