What You'll Need
Getting started with Pipeworks is quick. You do not need any technical experience, coding skills, or special software. All you need is:
- A web browser -- any modern browser works (Chrome, Firefox, Safari, Edge)
- An account with at least one business tool -- Stripe, Salesforce, Calendly, SendGrid, or any of the 20+ supported services
- An AI agent -- Claude, ChatGPT, or any agent that supports tool connections
That is it. The entire setup takes under five minutes.
Not sure which integration to start with? Pick the tool you use most often in your daily work. If you process payments, start with Stripe. If you manage a CRM, start with Salesforce. If you book appointments, start with Calendly.
The Seven Steps
Here is the full walkthrough, from creating your account to making your first AI-powered tool call.
Create Your Account
Head to app.pipeworks.ai and sign up. You have two options:
- Email and password -- Enter your email, choose a password, and you are in
- Sign in with Google -- One click and your account is created automatically
That is all there is to it. No credit card required, no approval process, no waiting. Your account is ready in about 30 seconds.
Your account is a platform identity that can belong to multiple workspaces. If you are invited to someone else's workspace later, you will use this same account to access it.
Create Your Workspace
After signing up, you will be guided through a short three-step wizard to create your workspace.
Name your workspace. Pick something that identifies your team or business -- like "Acme Marketing" or "Jane's Consulting." Pipeworks will suggest a URL-friendly slug based on the name, but you can customize it.
Pick your workspace type. Choose the option that best describes how you work:
- Small business -- You run a business and want AI to help with day-to-day operations
- Agency -- You manage tools and workflows for multiple clients
- Freelancer -- You work independently and want to streamline your own processes
- Enterprise -- You are part of a larger organization with team-level needs
See your recommended integrations. Based on your workspace type, Pipeworks suggests a starting set of integrations that are most popular for your use case. These are just suggestions -- you can connect any integration you want.
Your workspace type affects which integrations are recommended, but it does not limit what you can access. You can change your workspace type later or ignore the recommendations entirely.
Connect Your First Integration
From your workspace dashboard, go to the Integrations page. You will see the full catalog of available integrations. Click on the one you want to connect.
Each integration has its own connection process, but most follow the same pattern:
- API key integrations (Stripe, Klaviyo, SendGrid, Serper) -- Copy your API key from the service's dashboard and paste it into Pipeworks
- OAuth integrations (Google Analytics, Google Ads) -- Click "Connect" and sign in with your Google account. Pipeworks handles the rest.
- Application password integrations (WordPress, Elementor, Fluent Forms) -- Enter your site URL and an application password generated from your service
Once you enter your credentials and click Connect, Pipeworks encrypts everything and starts your integration.
Your credentials are encrypted with AES-256-GCM before they are stored. Pipeworks never saves plaintext credentials to disk. They are only decrypted at the moment your integration container starts, and they are passed securely through an encrypted environment file that is immediately deleted.
Set Permissions
This is where Pipeworks stands apart from simply sharing API keys with an AI. Before your agent starts using your tools, you can set precise, granular permissions.
Disable dangerous tools. Every tool is labeled as read, write, or delete. You can disable individual tools or entire categories. Most people start by disabling all delete tools so there is zero risk of the agent removing data.
Use tool profiles for quick setup. Instead of toggling tools one by one, you can apply a profile:
- Full access -- All tools enabled (read, write, and delete)
- Read-only -- Only tools that look up information, nothing that changes or removes data
- No-delete -- Read and write tools enabled, all delete tools disabled
Set per-user permissions. If you have team members in your workspace, you can assign different permissions to each person. Give your marketing intern read-only access while your operations manager gets full access. Each user gets their own scoped token, so their actions are tracked individually.
Start with read-only access. Let your AI agent answer questions and pull reports for a few days. Once you are confident, expand to write access. You can always tighten permissions later without disrupting anything.
Generate an API Key
Go to the Connect page in your workspace. This is where you create the credentials your AI agent will use to talk to Pipeworks.
Click Create API Key and give it a descriptive name, like "Claude Desktop - Personal" or "Production Agent." Then choose its scope:
- Select which integrations the key can access -- you do not have to grant access to everything
- Select which tools within each integration -- for even finer control
You will also choose the key type:
- Personal keys -- Tied to your user account. Actions are logged under your name. Best for individual use.
- Workspace keys -- Shared across the workspace. Actions are logged under "Pipeworks Service." Best for shared agents or automated systems.
After creating the key, copy it immediately. It is only shown once.
Named API keys replace the older "master key" approach. They are more secure because each key is scoped to specific integrations and tools, and every tool call is tracked with the key name in your activity log. You can have up to 25 active keys.
Connect Your AI
Now connect your AI agent to Pipeworks. The process depends on which agent you are using.
Claude.ai
The simplest option. Go to Claude.ai, open Settings, then Connectors. Add a new connector with this URL:
https://app.pipeworks.ai/tools
Claude will prompt you to sign in through Pipeworks using OAuth. Once authenticated, all your enabled integrations and their tools appear automatically inside Claude. This also syncs to Claude Desktop if you use it.
ChatGPT
In ChatGPT, go to Settings, then Apps. Create a new app and set the server URL to:
https://app.pipeworks.ai/tools
Choose OAuth as the authentication method. When you start a new chat, ChatGPT will prompt you to sign in through Pipeworks.
Claude Desktop (manual config)
If you prefer to use a named API key instead of OAuth, open your Claude Desktop configuration file and add:
{
"mcpServers": {
"pipeworks": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://app.pipeworks.ai/tools"]
}
}
}
Claude Desktop will prompt you to authenticate on first use. For a single integration, replace /tools with /tools/stripe (or whichever slug you want).
Other agents
Any agent that supports tool connections can work with Pipeworks. Point it to https://app.pipeworks.ai/tools and authenticate with either OAuth or a named API key passed as a Bearer token. Check the Connect page in your Pipeworks dashboard for agent-specific setup guides.
Use the aggregate endpoint (/tools) to give your agent access to all your enabled integrations at once. Use a per-integration endpoint (/tools/stripe) if you want to limit a specific agent to just one service.
Make Your First Tool Call
Open your AI agent and ask it something that uses your connected integration. Start simple -- use a read-only question so you can see the system work without changing any data.
If you connected Stripe:
- "List my 5 most recent payments"
- "How many customers do I have?"
- "Show me any failed charges from this week"
If you connected Salesforce:
- "How many open opportunities do I have?"
- "List my recent leads"
- "What deals are closing this month?"
If you connected Calendly:
- "What meetings do I have scheduled for tomorrow?"
- "How many appointments were booked last week?"
Your agent will use the Pipeworks tools to pull live data from your business service and respond with real, up-to-date information. You should see the tool call happen in real time -- most agents show which tool they are using and what data they received.
Every tool call is logged in your Pipeworks activity log with the timestamp, tool name, API key used, and response status. You always have full visibility into what your agent is doing.
What to Do Next
You are up and running. Here are some natural next steps:
Add more integrations. Now that you have the pattern down, connecting a second or third integration takes just a couple of minutes. Try combining a payment tool with a CRM, or a scheduling tool with email.
Enable write tools. Once you are comfortable with read-only access, expand your agent's permissions to include write tools. Now it can send emails, create invoices, update records, and take real actions on your behalf.
Invite your team. Add team members to your workspace and assign them roles. Owners and admins get full access; members only see integrations you assign to them, with per-user permission controls.
Create scoped API keys. As you build more workflows, create separate API keys for different purposes -- one for your personal Claude, one for a shared team agent, one for an automated system. Each key's activity is tracked independently.
Check the activity log in your workspace regularly. It shows every tool call, which key was used, and whether it succeeded. This is the best way to understand how your AI is interacting with your business tools and catch any issues early.
Frequently Asked Questions
How long does the whole setup take? Most people go from zero to their first tool call in under five minutes. Creating your account and workspace takes about a minute. Connecting an integration takes one to three minutes depending on how quickly you can find your API key. Connecting your AI agent takes another minute.
Do I need to install anything? No. Pipeworks runs entirely in the cloud. Your AI agent connects to it over the internet. There is nothing to download or install on your computer, unless you are using Claude Desktop (which has its own installer).
Can I try it without connecting to my live business data?
Yes. Many services offer test or sandbox modes. For example, Stripe provides test API keys (starting with sk_test_) that let you experiment with fake data before connecting your production account.
What happens if I revoke an API key? The key stops working within 30 seconds. Any agent using that key will no longer be able to make tool calls. Your integrations and data are not affected -- you can create a new key at any time.
Can multiple people use the same workspace? Absolutely. Invite team members from your workspace settings. Each person gets their own account, their own permissions, and their own activity log. You control exactly what each person can access.