Skip to main content
HubSpot keeps a record’s history as engagements: notes, emails, calls, meetings and tasks, each with a timestamp, an owner and the records it is associated with, plus attachments in its Files tool. Orbit keeps the same things: notes and tasks, logged activities for emails, calls, meetings and messages, and files on records. This page maps one to the other and gives the agent a plan it can follow end to end.
1

Connect both sides

Connect Orbit to your assistant (how) and approve crm:write, contacts:write, companies:write, deals:write, tasks:write, notes:write, activities:write and files:write. Then connect HubSpot’s MCP server next to it (it reads contacts, companies, deals and engagements), or export from HubSpot and attach the files.
2

Invite your team

HubSpot owners become Orbit members by matching email. Invite everyone first; whoever is not a member keeps their name on the records they wrote.
3

Set up the structure

Pipelines, stages and custom fields come first (Build your CRM). Ask the agent to show the mapping before it writes anything.
4

Run the migration

Records first, history second, files last, in the prompt below. Every row carries its HubSpot id, so the agent can stop and resume, and a second run updates instead of duplicating.

​
The prompt

“Migrate our HubSpot portal into Orbit. Use hubspot as the external_source everywhere and pass each HubSpot object id as the external_id.
  1. Companies: create_company for each HubSpot company with its name, domain, industry, size, owner (match by email to our members) and createdate as created_at.
  2. Contacts: import_contacts in batches of 500 with email, names, phone, job title, company, owner, createdate as created_at, and the HubSpot id as external_id.
  3. Deals: create_deal for each deal in the matching pipeline and stage, with amount, close date, owner and createdate; then add_deal_contact for its associated contacts (use contact_external_id).
  4. History: import_activities in batches of 500, one row per engagement, attached with contact_external_id / company_external_id / deal_external_id. Notes become note, calls call, meetings meeting, emails email, tasks task. Keep hs_timestamp as the time and the owner’s email as actor_email.
  5. Files: for every attachment, get its signed download URL from HubSpot and call import_file_from_url with the same record links and the HubSpot file id as external_id.
After each step, compare counts with HubSpot and list anything that did not import with the reason.”

​
Field mapping

The agent needs this when it reads engagements through HubSpot’s API or MCP server. Timestamps from HubSpot are Unix milliseconds; Orbit accepts them as they are or as ISO 8601.

​
Every engagement

​
Notes

​
Emails

​
Calls

​
Meetings

​
Tasks

​
Communications (SMS, WhatsApp, LinkedIn)

​
Large portals

A chat assistant handles a few thousand engagements; a portal with a hundred thousand emails and calls is a script’s job. scripts/maintenance/migrate-from-hubspot.ts in the Orbit repository does exactly the plan above against the CRM API with a HubSpot private app token and an Orbit API key (the CRM agent preset), in batches, and is safe to re-run. Point a coding agent at it, or run it yourself:

​
Check the result

Open a few contacts in Orbit: the timeline shows the migrated emails, calls, meetings and notes at the dates they happened, each marked with who did it and that it came from HubSpot. get_timeline reads the same feed for the agent. If a run went wrong, undo_activity_import removes everything that run created, records included only if they were created by that run (contacts stay; use the contact import’s own undo for those).