Skip to main content
Hive works on top of the systems your team already uses. You connect a tool once, Hive reads what you allow, and everything it notices or proposes stays tied to the source it came from. Nothing is migrated: your data stays where it is, and Hive reads it in place. To open Integrations, select your name at the bottom of the sidebar and choose Integrations, or search for it in the command palette (⌘ K). Before anything is connected, the empty Signals feed and the empty Business Brain also link here, and so does Connect your tools in a new Ask chat. The screen is headed Integrations and has two tabs:
  • Connectors — “Connectors let Hive read the tools you already run and carry out approved actions.”
  • Channels — “Ask Hive from Slack or Microsoft Teams. Signals and approvals stay in Hive.”

Recommendations from your website

Recommended for your firm, on the Connectors tab, suggests tools to connect first, each with its reason. Hive personalises it by reading your firm’s public website. If no one gave Hive the website during setup, the section says No recommendations yet: an Owner or Admin can type the website there (a full https:// address) and choose Find recommendations. Hive reads it in the background and the suggestions appear when it finishes. Other members see “Ask an owner or admin to add your firm’s website.”

Four ways to connect

Managed OAuth connectors

Sign in to a provider such as Xero, HubSpot, Gmail or GitHub. Hive holds the credentials in its vault. New accounts start on Act, and you can set each one to Draft or Read.

MCP servers

Connect an official vendor MCP server for a tool Hive has no regular integration for (Sentry, Confluence, Airtable, Granola and more) or any Model Context Protocol server by URL.

Incoming webhooks

Let any system push signed JSON events into Hive, directly or through Zapier or Make.

Slack and Teams channels

Ask Hive questions from Slack or Microsoft Teams with an @mention or a DM.
A connector brings data in and carries actions out to that tool. A channel is only a place to talk to Hive — it never delivers signals or approvals.

What Hive does with a connected tool

Once a tool is connected, Hive uses it in several ways at once:

Read, Draft or Act

Every connected account has its own level, and new accounts start on Act. In Ask (a chat with no agent picked) and Signals, the level decides what Hive does: on Read it tells you what it found and takes no action, on Draft it prepares the action for your one-click approval, and on Act it takes the action and tells you what it did, with undo where the action can be undone. Act never skips Hive’s safety checks: an action that cannot be undone, or one a check holds back, becomes an approval card. Bulk actions, saving a file to Drive and tools on MCP servers always ask first, and so does a plan you ask for with Act on this on a signal. Agents and workflows follow the same level for the actions selected for them, and even on Act wait for you when where an action sends did not come from you. See Accounts and autonomy.
To have Hive ask first in Ask and Signals, or only read, set the account to Draft or Read in its Accounts panel, which opens right after you connect.

Who can connect what

  • An Owner or Admin enables a provider for the workspace, simply by connecting it first.
  • After that, any member on a Basic, Pro or Max seat can connect their own account of that provider; a member on an Approver seat connects none. A member without a seat, or whose seat is suspended over the plan’s limit, can do nothing in the workspace, and Hive stops reading the accounts they connected until they have a seat again. Before a provider is enabled, members see: “A workspace administrator must enable this provider before you can connect a personal account.”
  • Every connection belongs to the member who made it. Owners and Admins see workspace connection metadata in the same Connectors grid as their own connections and the gallery. Personal credentials, configuration, scopes, webhook URLs and actions stay private to their owner.
  • Owners and Admins manage channels (Slack, Teams). Each member sets the Read, Draft or Act level of their own accounts, whatever their role; an Owner or Admin can cap a provider with a workspace ceiling, which only lowers it.

How fresh the data is

  • Right after you connect, Hive queues a first sync straight away and runs it in the background; the page updates on its own when it finishes.
  • After that, Hive’s always-on loop collects from every live connector hourly while the workspace is on a trial or a plan, including while a payment is overdue. Once the workspace is read-only (the trial ended without a plan, an unpaid invoice passed 14 days, the subscription ended, or Hive paused it), Hive stops collecting until a plan is active again or Hive resumes the workspace. See Trial and payments.
  • Some providers also forward realtime events to Hive, which are picked up as they arrive.
  • Gmail, Google Calendar, Google Drive, Outlook and Xero send only a “something changed” notice with no content. When one arrives, Hive re-reads that account’s mail or files within about a minute, from where its last read stopped, instead of waiting for the hourly run. A burst of notices for the same account becomes one read. Calendar events and Xero records are picked up by the next hourly run. A notice Hive cannot verify is ignored. For Outlook, Hive sets up these notices for your mailbox and calendar when you connect it and renews them every few days; see Outlook.
  • Owners and Admins see a countdown to the next hourly run: Next run in on the Connectors tab, and Next connector check in at the top of the Signals page.

What is in the catalog

Hive’s catalog covers accounting, payments, CRM, support, email, calendar, files, project tracking and developer tools. The tiles you see depend on how your Hive workspace is configured, and some providers are connect-only while their reads and actions are being built. The full list, with what each provider supports today, is in the integrations catalog.

Build your own connector

If the tool you need is not in the catalog:
  • If it has an MCP server, connect it by URL — see MCP servers.
  • If it can send webhooks (or you use Zapier or Make), point it at a signed incoming webhook.
The MCP servers and webhooks you created appear under Connected at the top of the Connectors tab, alongside the tools you connected.
Hive does not offer customer API keys or outgoing webhooks today. See the developer overview for what you can build against.

Connect an app

Step-by-step: connect your first tool.

Integrations catalog

Every provider and what it supports.

Health and disconnecting

Connection statuses and what disconnecting does.

Security overview

How credentials and data are protected.