AI Lead Generation & Outreach

Outbound that doesn't read like spray.

AI-powered B2B prospecting that researches each lead, drafts personalised emails referencing real signals (podcasts, posts, role moves), and only sends after you approve. Volume of automation, quality of hand-written.

You approve every send

GDPR-friendly

5-12% reply rate typical

Outbound pipeline · this week

Running

ICP build

stage 1

—

Prospect search

stage 2

412

Enrich + research

stage 3

284

Personalise draft

stage 4

284

Human approval

stage 5

247

Send

stage 6

247

Reply tracked

stage 7

Personalised draft · awaiting your approval

Draft 247/247

To: morgan.ohara@northway.io

Subject: Saw you spoke at SaaStock about renewals churn

Hi Morgan, Your SaaStock panel last month — the bit about post-renewal expansion — stuck with me. We've helped 3 RevOps teams build the same motion you described. Worth 15 minutes to swap notes? Rajesh

The outbound stack

Four moves an AI does well in outbound.

ICP-fit scoring

Score every prospect against your real ICP — industry, size, role, tech stack, signals. Filter out 80% that wouldn't have converted anyway.

Per-lead research

Pull each prospect's LinkedIn activity, podcast appearances, recent posts, role changes. Real signal, not boilerplate.

Personalised drafting

Draft based on real signal: 'saw your post on X', 'noticed your team's hiring for Y'. Reads like a human researched the lead — because the AI did.

Reply triage

Classify every reply as interested / not-now / out-of-office / unsubscribe. Route the right ones to your inbox, snooze the rest.

Where the human stays in the loop

The 90/10 rule: AI does 90%. You decide the 10% that matters.

Fully automated outbound that nobody reviews is how brands burn domain reputation. We design the workflow so AI handles volume, your salesperson handles approval — usually 10-15 minutes a day for 100+ touches.

AI handles

Sourcing from LinkedIn, Apollo, Cognism, public web
Enrichment with social signals + recent activity
Drafting the email with personalisation
Reply classification + auto-responses for OOO
Scheduling sends in your timezone window
Reporting + funnel metrics

You handle

ICP definition + voice/tone briefing (once)
10-minute daily review of drafted emails
One-click approve, edit, or kill per draft
Real conversations once replies come back
The actual sales meetings (still your strength)
Quarterly tuning of voice + ICP

Built on

Tools we wire together.

Apollo.ioCognismLinkedIn Sales NavHunter.ioClaySmartleadInstantlyLemlistClaude / GPT-4oHubSpotPipedriveSalesforcePostmark / SendGridCalendly

Honest limits

Three ways AI outbound burns reputation.

Spray-and-pray volume

Sending 1,000 cold emails/day with thin personalisation gets your domain blacklisted and your reply rate to zero. We cap volume per domain and ramp gradually. Quality of intent > volume of sends.

Fake handwritten claims

AI emails that pretend a real person hand-wrote them at 2am are unethical and increasingly illegal. We send transparently — automation doesn't have to mean deception.

GDPR ignorance

B2B email under GDPR in the EU has rules — legitimate interest basis, prominent opt-out, supplier data lawfulness. We design within these from day one. Sloppy outbound = ICO complaint risk.

Frequently asked

AI outbound questions.

Those tools handle volume but the personalisation is fundamentally template-based ('Hi {first_name}, I saw {company_name}…'). We add a research+drafting layer that actually reads each prospect's recent activity and writes a sentence that proves it. Reply rates typically move from 1-3% to 5-12%.

We don't replace your sending stack — we plug into it. If you have Smartlead or Instantly, we wire the AI research/drafting layer in front. If you don't, we recommend a stack to set up. You keep ownership of the deliverability infrastructure.

Supported but more carefully — LinkedIn rate limits and TOS are stricter. We don't use scraping or unofficial APIs. Approach is: AI does research, drafts the message, human sends from their authenticated session via Sales Nav. Lower volume, higher quality.

Properly personalised AI outbound to a well-defined ICP: 5-12% reply rate, 1-3% meeting-booked rate. We track and report monthly. Below those numbers usually means the ICP needs sharpening, not the AI.

B2B can be sent under 'legitimate interest' but must include: business context, prominent unsubscribe, no special-category data, lawful supplier data sourcing. We design within those rules and document the legal basis. ICO/DPC complaint risk minimised, not eliminated — outbound always carries some risk.

Build: €5,000–€12,000 setup. Ongoing: €600–€2,500/mo depending on volume, prospect data costs, and sending infrastructure. Often replaces an SDR salary at fraction of cost, plus your existing salesperson's time on real conversations.

Related AI services

Works alongside

Give us 50 dream prospects.

We'll show you what 50 personalised drafts look like — researched, drafted, ready to send. If you'd send them as-is, we've earned the conversation.