1
Intake
Spine polls or receives webhooks from Salesforce, HubSpot, Outreach, Gong, Gmail, LinkedIn, and Slack.
2
Load
Raw records are ingested into Spine staging with source metadata and ingestion timestamp.
3
Normalize
Records are mapped to the Spine entity schema: Lead, Account, Contact, Opportunity, Activity, and so on.
4
Schema Mapping
Cross-system matching links records by email domain, account identifier, fuzzy name, or phone number.
5
Context Assembly
Spine builds deal graphs by joining opportunities to contacts, activities, transcripts, and engagement events.
6
Projection
The assembled graph is filtered and shaped into domain views: My Pipeline, Forecast, Deal Reviews.
7
OODA
Twin observes the projected state, orients to deal health models, and decides whether to surface an alert.
8
Governance
Every projected view respects source-system permissions and domain guardrails before rendering.
9
Execution
When you invoke a capability, Spine routes the call to the correct connector with the right authorization.
10
State Update
Writeback responses update the local projection, trigger events, and refresh dependent views.