Integrations

Connect an AI to Gmail via OAuth and it can read inbox threads, write drafts, apply labels, and (with approval) send mail — all under user-level permissions. Most common automations built on Gmail: Lead Outreach, Meeting Prep, and reply drafters. Setup takes ~15 minutes after admin approval.

Connect an AI to Slack and it can read channel messages and threads, post messages and reactions, and DM users. The highest-ROI workflow is Support Triage. Rate limits and tenant policies are the main setup gotchas — coordinate with your workspace admin.

Connect an AI to HubSpot and it can read contacts, deals, companies, and engagement timelines, and write notes, tasks, and property updates. Core sales workflows built on it: Lead Outreach, Meeting Prep, Pipeline Hygiene. Custom properties need explicit mapping — it's the #1 setup gotcha.

Connect an AI to Salesforce for enterprise-grade CRM automation. The AI can read accounts, opportunities, contacts, and custom objects, and write tasks, notes, and field updates. Governor limits are the defining constraint — heavy queries need careful design.

Connect an AI to Notion via a native connector and it reads pages, blocks, and database rows, and writes new pages, updates, and comments. Core workflows built on it: RFP Drafter, Runbook Executor, and any knowledge-grounded automation that reads your internal docs. The per-page sharing model is the main gotcha — Notion parent-page permissions don't always propagate.

Connect an AI to Google Drive via a native connector and it can read Docs, Sheets, Slides, and PDFs in shared folders, and write new documents or edits. Core workflows: Invoice Reviewer (reads Finance folder), RFP Drafter (reads answer library), Report Generator. Folder-level scoping is the safest default — never grant 'all Drive.'

Connect an AI to Google Calendar via a native connector and it reads events, attendees, and free/busy data, and (with approval) creates events or RSVPs. Highest-ROI workflow: Meeting Prep — a one-page brief in the calendar invite 30 minutes before every external meeting.

Connect an AI to Google BigQuery via a service-account-backed connector and it runs SQL queries against authorized datasets, reads results, and writes summaries to Sheets, Slack, or Docs. Core workflows: Weekly Metrics Reporter, Revenue Variance Explainer, Churn Early-Warning. Cost monitoring is the biggest gotcha — chatty queries against large tables surface unexpected bills.

Connect an AI to Zendesk via Actions (OpenAPI) and it reads tickets, classifies them, drafts replies grounded in your macros and help center, and routes to the right queue. Ticketing is the highest-ROI integration for most support teams — a good Zendesk triage handles the bulk of repeatable work.

Airtable integrations go through Actions until a native connector lands. The AI reads bases, filters records, and writes new rows or updates. Airtable is the default lightweight database for many teams — great for content calendars, ops tables, vendor tracking. The highest-value workflows convert unstructured inputs into structured records.

AI integrations with Stripe focus almost entirely on read-only analysis and reporting. Write operations that move money or change subscriptions stay firmly human-controlled. The highest-ROI Stripe workflows are Revenue Variance Explainer, Subscription Health Monitor, and MRR Narrative Reporter. Setup via Actions until a native connector ships.

Jira integrates with an AI via Actions (Atlassian REST API) pending a native connector. The AI reads issues, creates tickets from unstructured input (customer feedback, support escalations), and generates status reports. Best workflows: Escalation Packager, Sprint Narrator, Bug Report Standardizer.