The services your clients can connect

On this page

Your clients don't all run on the same tools. One books patients in a scheduling app, another sells through Shopify, a third lives in HubSpot and QuickBooks. An automation agency needs to work in all of them, so Operate As supports thousands of services, from Gmail and Stripe to the niche app only one of your clients uses.

Here's what that covers, how clients connect each one, and how we make sure every service works before it's listed.

What your clients can connect

The catalog spans 26 categories, including:

• CRM and sales: HubSpot, Salesforce, Pipedrive and hundreds more
• Marketing, ads and SEO: email platforms, social scheduling, analytics and rank trackers
• Accounting and payments: QuickBooks, Xero, Stripe and the tools around them
• Online stores: Shopify and the apps that run inventory, shipping and reviews
• Projects, files and chat: Asana, Notion, Google Drive, Dropbox, Slack and Microsoft Teams
• Support, scheduling, HR, phone and SMS: Zendesk, Intercom, Calendly, Twilio and many more
• AI: OpenAI, Anthropic and hundreds of AI tools your clients already pay for

You can browse and search all of them in the integrations directory. Each service has its own page showing how a client connects it, what each permission means and every tool your agent gets.

Two ways to connect, both simple for clients

Every service connects in one of two ways, and the client page handles both:

• Sign in. For services like Google, Microsoft and Slack, the client presses Connect and signs in on the service's own screen. Their password goes to the service, never to you or us.
• Paste a key. For services that issue API keys, the client pastes one, with step-by-step help showing where to find it. Where a service allows it, they can make a restricted key that reaches only what you need.

Either way, every permission is written as a plain sentence, so clients know exactly what they're sharing.

Every tool, plus any endpoint

Each service comes with ready-made tools for its API, tens of thousands across the catalog. Your agent doesn't need to load them all: over MCP it uses search_tools to find the right one from a few words, like "read sheet rows" or "list open invoices".

When a job needs something the tools don't cover, the request tool calls any endpoint of the service's own API, with the client's access attached on our server. If the service's API can do it, your agent can do it.

Checked against the live service

A catalog this size is only useful if it works. So we don't hand-write wrappers that drift out of date. Each service's tools come from the API description the service publishes, or from its own documentation, and every service is checked against its live API before it's listed.

The check confirms the parts that have to be right for every call: the address the API lives at, where the client's key or sign-in goes, and that the service's tools exist. A service that fails isn't listed until it's fixed. Each service's result and the day it was checked are shown in the directory, and the full method is in our docs.

Missing a service?

If a client uses something you can't find, search for it in the directory and send us a request from there. We'd love to hear which services matter to your clients.