CRM integrations

Sync leads, conversations and positive replies to HubSpot, Pipedrive or Salesforce, and feed GraphKit lists from HubSpot lists.

Connect your CRM once per workspace and GraphKit keeps it current. Nobody copies replies into the CRM by hand:

In GraphKitIn the CRM
A lead you message, or every lead (optional)A contact (HubSpot), a person (Pipedrive), or a lead or the existing contact (Salesforce). Its company is created and linked where the CRM has companies.
Each message sent or received, on any channelA note (HubSpot, Pipedrive) or a completed task (Salesforce) on that record, logged once
A reply classified interested, by AI or by a teammateA deal in the pipeline and stage you choose (HubSpot, Pipedrive), plus a note with the reply. Salesforce gets a completed task.
A HubSpot list you linkA GraphKit lead list you can use in any campaign, re-imported every hour

Connect

Open Integrations → CRM. You need a role that can manage integrations (client admin or an agency role).

  • HubSpot: choose Connect HubSpot, pick the HubSpot account and approve. GraphKit asks for contacts, companies, deals and lists access.
  • Pipedrive: choose Connect Pipedrive (OAuth), or Use an API token and paste the token from Settings → Personal preferences → API. An API token acts as the person who created it.
  • Salesforce: choose Connect Salesforce and sign in. GraphKit uses the OAuth web server flow with PKCE and asks for api and refresh_token.

Credentials are encrypted at rest and bound to the connection. Access tokens are refreshed automatically. If the CRM refuses GraphKit's credentials (the app was uninstalled, or the refresh token expired), the connection shows Needs attention and sync pauses. Nothing is lost: choose Reconnect and the queue continues. A workspace has one connection per CRM; reconnecting replaces its credentials.

Sync settings

SettingDefaultWhat it does
Create and update contactsOnOff: activity is only logged on records that already exist (matched by email or LinkedIn URL); other leads are skipped.
Push every leadOffAlso sends every lead created or edited in the workspace, including imports. Off: only leads with activity.
Log each messageOnSent and received messages, once each.
Open a deal on a positive replyOnChoose the pipeline and stage. If the lead already has a deal from GraphKit, it moves to that stage instead of creating a second one. Until you choose a stage, the reply is logged as a note only.
Include private conversationsOffConversations the inbox keeps private (see Inbox) stay out of the CRM, where everyone with CRM access could read them.

Suppressed leads are never sent to the CRM, whatever the settings. Their changes show as Skipped in the sync log.

Field mapping

Field mapping lists every GraphKit lead field, including your custom fields, next to a property picker filled live from the CRM. Unmapped fields are not sent. Suggest from HubSpot (or Pipedrive, Salesforce) fills empty rows by matching property names and labels; review before saving.

  • Map LinkedIn profile URL to a CRM property. GraphKit then matches existing records by LinkedIn URL when the email is unknown, and you can import lists.
  • GraphKit never overwrites the email on an existing record: it is how records are matched.
  • Pipedrive custom fields are written to the person's custom fields; Pipedrive's update never replaces the emails and phones your team added.
  • Salesforce requires a last name and a company on a lead. When the lead has none, GraphKit sends the full name and "Unknown".

How sync works

Every minute GraphKit looks for changes since the last run and queues one operation per change. Each operation has a key (for example, the message id), so a change is never queued twice and a record is never created twice: GraphKit remembers which CRM record each lead became, and searches by email and LinkedIn URL before creating one.

Operations for one connection are sent one at a time. A failed operation is retried with growing delays (1, 3, 9, 27 minutes and so on, with jitter, up to 12 hours apart) for up to 8 attempts, spread over as much as 18 hours. A CRM's Retry-After is respected. After the last attempt the operation is Failed and stays in the Sync log. Fix the cause, often a mapping the CRM rejects, then choose Replay. If a record GraphKit created was deleted in the CRM, the next sync creates it again.

Sync now looks for changes immediately instead of waiting for the next minute.

HubSpot lists as a lead source

  1. Map LinkedIn profile URL in Field mapping.
  2. Open Lists, search your HubSpot lists and choose Feed a GraphKit list.
  3. Pick a new or existing GraphKit list and whether to re-import every hour.

Members with a LinkedIn URL become leads (source integration), matched to existing leads instead of duplicated, and are linked to their HubSpot contact so later syncs update it. Members without a LinkedIn URL are skipped and counted on the list. Leads that leave the HubSpot list stay in GraphKit. Use the GraphKit list in a campaign like any other list.

Pipedrive filters and Salesforce list views cannot feed GraphKit lists yet.

For operators: register the OAuth apps

Each CRM's Connect button appears only when its OAuth app is configured. API_URL is the public origin of the GraphKit API.

CRMWhereRedirect URIScopesEnvironment
HubSpotA HubSpot public app (developer account → Apps)API_URL/api/app/oauth/hubspot/callbackoauth crm.objects.contacts.read crm.objects.contacts.write crm.objects.companies.read crm.objects.companies.write crm.objects.deals.read crm.objects.deals.write crm.lists.read (mark them required)HUBSPOT_CLIENT_ID, HUBSPOT_CLIENT_SECRET
PipedrivePipedrive Developer Hub → create an appAPI_URL/api/app/oauth/pipedrive/callbackTick Base, Contacts: full, Deals: full, Activities: full in the app (Pipedrive takes scopes from the app, not the URL)PIPEDRIVE_CLIENT_ID, PIPEDRIVE_CLIENT_SECRET
SalesforceAn External Client App (or Connected App) with OAuth enabled, PKCE required, secret required for the web server flowAPI_URL/api/app/oauth/salesforce/callbackapi, refresh_tokenSALESFORCE_CLIENT_ID, SALESFORCE_CLIENT_SECRET, optional SALESFORCE_LOGIN_URL (https://test.salesforce.com for sandboxes)

API versions GraphKit calls: HubSpot CRM 2026-09 and OAuth v3, Pipedrive API v2 (v1 for notes and the current user), Salesforce REST v68.0.

Permissions

ActionWho
See connections and the sync logAnyone who can read campaigns
Connect, change settings and mapping, link lists, replayClient admins and agency roles (integrations)

On this page