01
Pull target accounts
Apollo.io search by ICP filters (industry, size, geography, tech stack signals). Pushes new accounts to Airtable with dedup by domain. ~150 candidates per Monday.
How the pipeline runs
Monday 6am the workflow runs end-to-end. By 8am every account is rescored, every new trigger event is logged, every outreach draft is regenerated.
01
Apollo.io search by ICP filters (industry, size, geography, tech stack signals). Pushes new accounts to Airtable with dedup by domain. ~150 candidates per Monday.
02
Clearbit + LinkedIn Sales Nav for employee count, revenue band, recent hires. Stored as JSON on the Airtable record.
03
Five n8n branches in parallel: funding rounds (Crunchbase API), exec hires (LinkedIn + Wappalyzer signals), tech stack changes (BuiltWith webhook), conference signals (RSS from key events), news mentions (Diffbot).
04
Apollo people-search for titles matching your ICP. Cross-referenced with HubSpot for prior-touch flags. Result: a ranked DM list per account with budget-owner annotations.
05
Composite score = 0.40 ICP fit + 0.35 trigger density (weighted by severity + recency) + 0.25 buying-stage signals. Rescored every run.
06
Claude 3.5 Sonnet generates one personalized opener per account, anchored to the highest-severity trigger + the budget-owner DM. Tone matched to your prior sales emails (one-time training corpus, ~30 emails).
07
New accounts created in HubSpot with all enrichment, trigger annotations as note attachments, draft outreach as sequence-ready emails. Owner assignment by territory.
08
Slack DM to the AE with their top three accounts for the week, each linked back to Atlas for full context.
n8n
Workflow orchestration
Apollo.io
Account + DM data
HubSpot
CRM of record
Airtable
Trigger event store
Clearbit
Firmographic enrichment
Wappalyzer
Tech stack monitoring
Claude 3.5
Outreach drafts
Slack
AE notifications
Demo: data is sample, workflow is mapped but not running live. Production: real n8n instance on the same droplet, $0/mo additional, every node logged for audit.