Each contact carries identity data, a computed engagement score, tags, custom fields, and consent preferences. Companies aggregate their contacts into an account-level view. Duplicates are merged. It is the behaviour graph the rest of the platform draws from, not a place to run pipelines.
Open a contact and you see the whole history: every page viewed, email opened, form submitted, chat message, and Knowledge Base search. Each activity is timestamped and attributed to the app that logged it. Ask for a summary and the AI writes one from the timeline on demand.
Activities log every interaction across every channel, queryable in plain English. Ask "who are our most engaged contacts at Acme that we haven't emailed in thirty days," and the AI queries the CRM directly and returns the list with context. It flags anomalies too, like a contact whose engagement dropped sharply.
This is a marketing-native CRM, not a sales-ops one. There are no deal stages, forecasting, quotas, territories, or commission math — for those, Salesforce, HubSpot Sales Hub, or Pipedrive is the right tool. There is no dialer, call recording, or sales sequences.
Custom fields are relatively simple; deeply nested or relational data models are not supported. And there is no contact enrichment from outside providers like Clearbit or ZoomInfo today, though that can be added by webhook.
Ask for early access. Import a few contacts or let a form create them, then ask the AI a question about them in plain English and see what it already knows.