Connecting business systems

A connection syncs records and documents from a system you already run into your knowledge, read-only. Admins and Owners can add connections from Settings → Connectors (Add a connector) or Knowledge → Sources → Connectors.

What you can connect

ERP: Epicor Kinetic (customers, parts, orders, quotes, invoices, custom BAQs), SAP S/4HANA (customers, materials, sales orders, deliveries, invoices), Dynamics 365 Business Central (customers, vendors, items, orders, invoices, shipments, purchase orders).
Product and data: Akeneo PIM (products, images, datasheets), BigCommerce (products, images, store pages, customers, orders), Directus (collections as records).
CRM and marketing: HubSpot (contacts, companies, deals, tickets, custom objects), Salesforce (contacts, leads, accounts, opportunities, cases, knowledge articles), Dynamics 365 (accounts, contacts, leads, opportunities, cases, knowledge articles), ActiveCampaign (contacts, accounts, deals, lists, automations, campaign performance).
Support and work tracking: Zendesk (tickets, comments, Help Center articles), Intercom (conversations, Help Center collections, contacts), Jira (issues and comments), ClickUp (tasks and Docs).
Files: Google Drive, Microsoft 365 (SharePoint document libraries), Dropbox. Nothing syncs until you pick folders.
Chat and analytics: Slack (public channels you turn on under Ingest channel knowledge; the bot must be in them), Google Chat (opt-in spaces), Google Analytics (GA4 metrics).
Meetings: Records Labs Notetaker, which joins Google Meet, Zoom, and Microsoft Teams calls and adds the transcript, a summary, and action items.
Personal, per member: Gmail, Outlook, Google Calendar, Google Contacts.
Custom API: any REST API, see below.

Microsoft Teams channels are listed under Coming soon.

Setting up a connection

Click Add a connector and pick the system.
Give it a Connection name.
Sign in or enter credentials. OAuth systems (for example HubSpot, Salesforce, Zendesk, Intercom, Google, Microsoft, Dropbox, ClickUp, and Jira) show Authorize {name} and send you to the provider. Key-based systems ask for their own fields, for example Epicor's Kinetic server URL, Company ID, integration account username and password, and API key.
Click Connect & Scan (or Connect).

For record systems, a Set up {name} flow opens next:

Choose Data — the objects your credentials can read. Click Enable recommended or pick them one by one.
Live Lookups (Epicor, SAP, Business Central only) — which objects answers may look up live.
Access & Trust — defaults applied to everything this connection brings in.
Review & Start — a receipt showing Access, Syncing, First Sync, Schedule, and Change Policy, which sets whether New objects and New fields Sync automatically or Hold for review, and whether to Notify admins when the schema changes.

Click Start syncing. Nothing syncs until at least one object or location is chosen.

Settings every connection has

Open a connection from Settings → Connectors to see its cards:

Connection — credentials, with Reconnect {provider} if authorization lapses.
Permissions — the default audience: Everyone in the organization, Specific people, or Just me. New records stay out of answers until a connection has a default audience.
Sources — each object or location, with Enable or Disable and an Override editor for category, team, tags, visibility, trust, wording, and sync frequency. New defaults reach records already synced only when you ask: under Saved defaults, Apply to on-default records updates records without an override, and Force reset all overwrites every record.
Sync — schedule and scope, with Sync now and Last sync.
Remove connection.

The Access & Trust step sets Who can see this?, Trust (Low, Medium, High), Stays accurate for, How the AI may use the wording, Owning team, Category, and Tags. Sensitivity and External Channels are set per record in the Library; synced records stay internal and hidden from customers until someone opens them up. A source override wins over connection defaults, and a record override wins over both.

Sync schedules

Sync frequency offers Manual (only when you press Sync now), Hourly, Daily, and Weekly. New connections start on Manual, and only organization admins can run a manual sync.

Each plan includes a fixed number of business-system connections. At the limit, Records Labs asks you to remove a connection, add one from Settings → Plan & usage, or upgrade. Websites and uploads never count.

Status, pause, and removal

Connections show Connected, Syncing…, Draft, Authorization required, Reconnect needed, Paused, or Needs attention. Knowledge → Sources also shows Synced — waiting for audience when a connection has no default audience yet.

Remove connection asks what to do with synced data: Keep records active (searchable and citable, but no longer syncing), Move records to Trash (restorable from Library → Trash), or Purge records.

Records that appear in more than one system

A record with the same type and identifier in two systems becomes one record with all fields combined. When sources disagree on a field, the higher-ranked system wins (ERP, then PIM, commerce, CRM, work trackers, support, imports, crawls). A disagreement from an equal or lower-ranked system keeps the current value and goes to Knowledge → Review. Matching on other fields is only proposed; you approve every match. A record shared by two sources is not deleted when one drops it.

The Custom API connector

Custom API connects any REST API without code. It is read-only (GET requests only), requires HTTPS, and stays on the host of the API base URL you enter. Set up: API base URL, Auth kind (bearer, basic, header, or query), Header / parameter name, and API credential.

Then open the Connector builder card. Describe What should it pull?, optionally paste an API docs excerpt and a Sample endpoint path, and click Draft connector. Check the Live preview: what a sync would pull, then Revise or Approve & create sources. A connector can have up to 12 collections. Each collection pulls up to 5,000 items per sync, and a larger collection picks up where it left off on the next sync.