CRM Integrations — Salesforce, HubSpot

Connecting your CRM gives every Omniflow conversation real customer context: account tier, lifetime value, open deals, recent renewals. It also writes back — conversations and tickets become activities on the CRM record so your sales and CS teams see the full picture.

Supported CRMs

CRM	Bidirectional?	Tools exposed
Salesforce (Sales Cloud, Service Cloud)	Yes	`lookup_account`, `lookup_contact`, `update_case`, `create_task`
HubSpot	Yes	`lookup_contact`, `update_deal`, `create_engagement`
Pipedrive	Read-only	`lookup_person`, `lookup_deal`
Zendesk Sell	Read-only	`lookup_lead`, `lookup_deal`

Other CRMs can be wired up via Custom webhooks.

Connect Salesforce

Open Integrations

Settings → Integrations → Salesforce → Connect.

Pick environment

Production or sandbox. Picking sandbox lets you test the connector without affecting live data.

Authorize

You’ll be redirected to Salesforce. The connected user becomes the integration owner and the actor for write-backs — use a service account, not a personal user.

Pick objects

Object	What it does
Account	Maps to Omniflow Companies.
Contact	Maps to Omniflow Contacts.
Lead	Optional — maps to Contacts.
Case	Maps to Omniflow Tickets.
Opportunity	Read-only context on contact / company records.

Map fields

For each object, map source fields to Omniflow fields. Custom fields are supported — Omniflow creates matching columns on Contacts / Companies.

Choose write-back behavior

Event	What gets written back
Conversation closed	A Task or completed Activity on the Contact + linked Case.
Ticket created	A Case (Service Cloud only).
Ticket resolved	Case status update + resolution comment.

Save and run initial sync

Initial sync can take 5-60 minutes depending on volume. Subsequent syncs are incremental and run every ~5 minutes.

Connect HubSpot

The flow is the same — pick objects (Contacts, Companies, Deals, Tickets), map fields, choose write-back. HubSpot’s free tier limits some scopes; the connector tells you up-front if your plan blocks an option.

Using CRM data in conversations

Once synced, every Omniflow contact has a CRM panel:

Account tier, lifetime value, deal stage.
Open opportunities.
Last 5 CRM activities.
Direct link back to the CRM record.

Agents see this on every conversation; the AI agent has it as tool-callable context.

Using CRM data in agents

Every CRM connector exposes lookup tools the agent can call:


Agent: "Let me pull up your account..."
Tool call: salesforce_lookup_contact(email="maya@acme.com")
Tool result: {tier: "gold", arr: 25000, owner: "alex@acme.com"}
Agent: "I see you're on our gold plan — let me get this resolved quickly."

Tools respect Salesforce field-level security. The integration user’s permissions limit what the agent can read and write.

Common pitfalls

Pitfall	Fix
Duplicate contacts after sync	Run Contacts → Merge duplicates after the initial sync.
Custom fields not appearing	Re-run Sync schema after adding fields in Salesforce.
Write-backs failing silently	Check the integration’s audit log — usually a missing scope or field permission.
Slow initial sync	Throttle the connector to off-hours under Settings → Integration schedules.

Don’t grant the integration the System Administrator profile. Use a dedicated profile with object-level read + the specific writes you want — overly broad permissions are an audit risk.

Open in Omniflow

If you want to…	Go to
Use the CRM as an agent tool	Tools & Actions
Build routing rules from CRM data	Routing Rules
Sync from a CRM not on this list	Custom API & Webhooks