Sales Ops Engine

AI outreach pipeline with personalised video, built on n8n

AI AutomationWorkflow OrchestrationSales OpsIntegrations

At a glance

Role: Design and engineering
Stack: n8n · GPT-4o · D-ID · HubSpot · Airtable · Gmail · Slack · WhatsApp

~$3K/momanual BDR work replaced

Project details

A client's sales team was spending BDR hours on work that follows a pattern: research a lead, find something worth mentioning, write a message around it, send it. We built an n8n orchestration that does that work end to end, including a short personalised video for every lead, and keeps a person in charge of what actually goes out.

Four tiers

Lead intelligence: GPT-4o reads each lead's signals from HubSpot and Airtable and picks a specific hook to open with. Personalisation: that hook drives two outputs in parallel, copy written for that lead and a custom AI-avatar video generated through D-ID. Review: everything lands in an Airtable gate, where a person reviews it before anything is sent. Delivery: approved messages go out through Gmail, Slack and WhatsApp, with sends throttled.

Why there is a human gate

Automated outreach fails in public when a generated line misreads a lead. The review gate means every message and video that reaches a prospect has been seen by a person first. The tiers are modular, so a channel, a model or the video provider can be swapped without rebuilding the rest.

How a lead moves through the engine

Signal — a lead in HubSpot or Airtable is picked up with its CRM signals
Hook — GPT-4o analyses the signals and chooses an angle for that lead
Write — outreach copy is drafted around the hook
Record — D-ID generates an AI-avatar video for the same lead
Review — copy and video land in the Airtable gate for a person to check
Send — approved outreach goes out through Gmail, Slack or WhatsApp, throttled

Where it fits

The engine works alongside the client's existing tools rather than replacing them: HubSpot and Airtable stay the source of truth for leads and for review, and messages go out through Gmail, Slack and WhatsApp.

The engine replaces an estimated $3,000 a month of manual BDR work while lifting response rates.