An AI outbound sales playbook is a team procedure for turning account information into approved outreach and a documented sales handoff. AI can help research, draft, and sort replies; an operator checks contact eligibility, message claims, and the next action before customer-facing work begins.
What job should the campaign brief define?
Name the target market, offer, permitted data sources, exclusions, owner, and deliverable. In Dureach, this can be assigned as a business job to a named specialist in an operator-reviewed AI workforce. Request a reviewable account list, message draft, reply rules, and CRM handoff rather than a send-volume target.
| Brief field | Example | Approval question |
|---|---|---|
| Account fit | Importers of a product category in a chosen market | Is this the right buyer role? |
| Signals | Shipment or customs activity with source and date | Does the record support this inference? |
| Exclusions | Existing customers, opt-outs, restricted uses or jurisdictions | May this contact be approached? |
| Output | Account note, draft sequence, owner-ready CRM note | Who reviews each artifact? |
How does information become a contact decision?
Use the same sequence for trade and other B2B campaigns:
Signals -> Leads -> Enrichment -> Decision -> Outreach -> PipelineFor example, a shipment record may identify a possible importer. Check the entity name, domain, country, trade role, and any CRM match before selecting a contact. A consignee or freight forwarder may not be the buyer. Record the source and uncertainty; enrich only fields needed for fit, routing, or a lawful contact decision. Treat AI findings as candidate data. Validate and deduplicate them against authoritative records, and send conflicts to an operator.
Use these editorial decision labels, not product UI statuses:
| Decision | Reason | Next action |
|---|---|---|
| Review for outreach | information and identity are credible | Check eligibility and approve the message |
| Research further | Role or source is uncertain | Verify before contact |
| Route internally | An owner or open opportunity exists | Share context with that owner |
| Suppress | Opt-out, restriction, or disqualifying risk | Do not contact; retain suppression as policy requires |
How should the team approve a launch?
Draft a short message from an observation the source supports:
Supported observation -> possible relevance -> practical offer -> small next stepAn operator checks the audience, source provenance, sender, claims, personalization, follow-up timing, and reply routing before launch. Check consent metadata, opt-outs, suppression, restricted uses and jurisdictions, trade-data licensing, tracking, retention, rate limits, and international data-use rules. A shipment record is information of recorded trade activity; it does not prove current demand or permission to contact a person. Confirm sender authentication and list quality, then set campaign-specific pause conditions for bounces, complaints, and poor fit. Do not promise a universal send rate or meeting result.
| Stage | Review point | Output |
|---|---|---|
| Research | Source, identity, duplicates, uncertainty | source-backed account list |
| Launch | Eligibility, approved claims, sender and pause rules | Approved sequence |
| Replies | Opt-outs, objections, referrals, owner routing | Reviewed next action |
| Pipeline | information, thread summary, caveats, ownership | CRM-ready handoff |
Where does Deep Research fit?
Use Deep Research for a broad account or market question before building a campaign. In Dureach it is a mode inside /app/copilot, with durable plans and artifacts in /app/ai-projects:
Prompt -> Clarify -> Plan -> Approve -> Research -> Validate -> Match -> Deduplicate -> Artifact -> ReviewClarification is optional and limited to material gaps. Execution begins after plan approval. The server-owned run survives browser closure and shows observable work and research summaries, never hidden reasoning. Validate candidate outputs, preserve provenance and replayable events, keep logs free of secrets, and leave authoritative records canonical until reviewed changes are approved.
How should replies reach the right owner?
Define actions before sending: interest goes to a sales owner, a referral needs identity review, an objection needs an approved response, and an unsubscribe must enter suppression. Route ambiguous or sensitive replies to a person. Do not let a draft reply make an unapproved pricing, legal, or contractual commitment. The handoff should include account information, contact role, campaign source, reply summary, caveats, owner, and next action. Review qualified conversations and CRM follow-through against the team’s own baseline, not a promised benchmark.
What should the team do next?
Write the brief, identify an owner for each approval, and review one source-backed account before expanding the campaign. To assign the work in Dureach, hire an AI agent with the brief and required deliverables. For the specialist’s role, see the AI outbound sales agent guide.
What are common questions about an AI outbound playbook?
What is an AI outbound sales playbook?
It is a documented sequence for research, eligibility review, approved outreach, reply handling, and CRM handoff, with named owners and approval points.
Does trade records prove buying intent?
No. It shows recorded activity that can justify research. Confirm the entity and role before drawing a sales conclusion.
Should AI send every drafted message?
No. Approve the audience, information, claims, sender, and follow-up rules before customer-facing outreach.
Which contacts should be excluded?
Exclude opt-outs and contacts barred by your consent, suppression, restricted-use, jurisdiction, or client rules. Review uncertain matches.
What happens when a prospect unsubscribes?
Stop follow-up and record the suppression according to the applicable policy. Review any queued actions that could contact the person again.
What belongs in the CRM handoff?
Include the account reason, source, contact role, conversation summary, caveats, assigned owner, and agreed next step.