Turn OpenClaw into an outbound SDR

OpenClaw is a self-hosted personal assistant; on its own it doesn't know how to prospect or write a cold email. Give it the right skills and it can run the full motion: research an account, draft a first touch built on one real signal, queue a follow-up cadence that stops on reply, triage what comes back, and book the meeting on your calendar.

What it can't do: guarantee replies or pipeline, verify a prospect's consent, or send safely without you configuring a warmed sender. It drafts and schedules — you keep a human on the first batches until you trust the voice. Think of it as a senior-SDR playbook an agent follows, not a hands-off autopilot.

The AI SDR Pack is six skills built to OpenClaw's AgentSkills spec, designed to hand off to each other:

• prospect-research — enrich a lead from a domain or name: what they do, recent buying signals, and the single best angle to lead with.
• cold-open — draft a personalized, non-templated first touch from that research (one observation, one hypothesis, one soft ask).
• followup-sequence — generate and schedule a 4-touch cadence that stops the instant they reply.
• reply-triage — classify inbound (interested / not now / objection / referral / OOO / opt-out) and draft the next move.
• meeting-booker — offer real times in the prospect's timezone, confirm, and drop a calendar hold with an agenda.
• pipeline-log — one running deal log every other skill reads and writes, so the assistant remembers each deal.

The chain: research → cold-open → followup-sequence → reply-triage → meeting-booker, with pipeline-log as shared memory.

The skills describe *what to do*; they use the tools and channels you connect to OpenClaw. To get the full motion working:

• Email channel (cold-open, followup-sequence, reply-triage): the sender you send from — ideally a warmed secondary domain — with read access so triage sees replies and stops cadences.
• Calendar (meeting-booker): Google Calendar or your provider with write access, plus a default Meet/Zoom link.
• Web search / page fetch (prospect-research): a search backend so research isn't guessing.
• Pipeline store (pipeline-log): a local pipeline.json by default, or point it at a Google Sheet, Airtable, HubSpot, or Pipedrive.
• Enrichment API (optional): Apollo/Clearbit for firmographics and verified emails; without one, research works from the public web.

The pack ships an openclaw.json.example to start from — you adapt names to your setup.

OpenClaw has a free public skill registry (ClawHub), and you can absolutely assemble an outbound stack à la carte from free skills — that path is legitimate and costs nothing. What you're buying here is the curated, assembled layer: six skills chosen to cover one job end-to-end, written to hand off to each other, tuned so the cold opens don't read like a mail-merge, and shipped with a setup guide and example config.

The do-it-yourself alternative is finding, vetting, and wiring individual skills together, then discovering after your first campaign which gaps you missed — reply triage, a shared pipeline log, a follow-up cadence that actually stops on reply. The pack is the "it just works for outbound" shortcut, backed by a 14-day refund.

No pack guarantees pipeline — reply and meeting rates depend on your offer, list quality, and email deliverability, none of which a skill controls. Two things stay your responsibility.

Compliance: cold outbound is regulated (CAN-SPAM in the US; GDPR/PECR in the EU/UK). The skills honor opt-outs immediately and avoid tracking links on the first touch, but they won't verify consent or guarantee legal compliance for your jurisdiction. Know your obligations.

The human loop: the skills draft and schedule, but you should review the first batches — especially positive and hostile replies — until you trust the voice. Used that way, an OpenClaw SDR setup removes the research-and-typing grind, not your judgment.