In the video, the AI assistant:
- Researches current minimum wages for all US states online.
- Creates two apps:
US Minimum Wagewith one record per state using data from the online research.Compliance Teamswith four team records.
- Relates each state to its responsible compliance team based on its region.
All from one prompt. Welcome to the Tape MCP Server public beta.
What you can do
Once connected, your AI assistant like ChatGPT or Claude can work directly with Tape.
- Ask about your data
Find overdue tasks, inactive deals, records matching specific conditions, or information spread across different apps. - Create and update data
Create records, update fields, add comments, mention users, or make changes across many records. - Build apps and views
Describe what you need and let AI create apps, fields, records, relations, and views. - Work with automations
Inspect existing automations, create new ones, analyze runs, or run an automation on demand. - Analyze your Tape setup
Check field dependencies, calculation fields, relations, automations, and other parts of your organization. - Create reports and tools
Combine data from different apps into Excel reports, dashboards, relationship maps, and other outputs.
We created videos and complete prompts for some of these examples.
See Tape MCP Server use case library ➔
What is MCP?
MCP stands for Model Context Protocol. It is a standard that lets AI assistants like ChatGPT or Claude connect to applications like Tape, find information, and take action for you.
Think of it like USB-C for AI. One plug works with devices from different brands. MCP does the same for AI assistants and applications.
MCP was introduced by Anthropic in November 2024 to solve a simple problem. Without a common standard, AI companies would have to build and maintain separate integrations for thousands of different applications.
MCP shifts that work to the applications themselves. Instead of Anthropic, OpenAI, or another AI provider building a custom Tape integration, Tape provides one MCP Server that different MCP-compatible AI assistants can connect to.
In December 2025, Anthropic contributed MCP to the Agentic AI Foundation under the Linux Foundation, giving the standard a neutral home for its continued development.
How MCP works
Every MCP setup has four core parts:
- AI assistant
Where you make your request, such as ChatGPT, Claude, Cursor, or VS Code. The AI understands what you want. - MCP client
The connection between the AI assistant and an MCP server. It sends requests and receives responses. - MCP server
The application side. In this case, Tape. It provides the available tools and handles requests. - Tools
The actions available to the AI, such as finding records, updating fields, creating apps, or working with automations.
For example, you ask “Create a new task in Tape.”
· The AI understands the request.
· The AI picks the right tool.
· The MCP client sends the request to the Tape MCP Server.
You ➔ AI assistant ➔ MCP client ➔ Tape MCP Server
MCP vs. API FAQs
Picture a restaurant. The API is the kitchen. It contains everything needed to prepare a meal, but you need to know exactly how the kitchen works to order from it directly.
MCP is the menu. It shows the AI assistant what is available and how to order it. Instead of understanding everything happening in the kitchen, the assistant can simply choose the right item from the menu.
Is MCP the same as an API?
No. The Tape API lets developers interact with Tape directly through code. MCP adds a standard layer for AI assistants to discover and use available Tape tools. The Tape MCP Server uses Tape's APIs underneath.Does MCP replace the Tape API?
No. Use the Tape API when you want to build a direct integration with code and control exactly what happens. Use the Tape MCP Server when you want an AI assistant to understand your request and decide which Tape tool to use.What's the difference between MCP and Tape Automations?
Tape Automations are deterministic: you define the trigger, conditions, and actions in advance. MCP is AI-driven: you describe what you want, and the AI decides how to use the available Tape tools. If something needs to run the same way every time, use Tape Automations.Do I need to understand APIs to use the Tape MCP Server?
No. You describe what you want in natural language, and the AI handles the technical interaction for you.Before you start
A few things are important to keep in mind:
- AI is not deterministic. The same prompt can produce different results, miss information, or interpret data differently. If something needs to run the same way every time, use Tape Automations instead.
- AI can make mistakes. Results are not guaranteed to be accurate or complete. Always verify important results and changes directly in Tape.
- Changes can trigger automations and webhooks. These may update or delete other records.
- There is no delete tool, but data can still be changed or deleted. AI can overwrite values in records, and automations it triggers or runs can include delete actions.
- Changes cannot be undone through MCP. Review important changes before running them.
Get started with Tape in your AI assistant
Your AI assistant only gets access to the parts of Tape you allow.
Connect Tape using a Personal Access Token. You choose which workspaces, apps, and capabilities it can access, and a token can never have more access than you have in Tape.
We recommend starting with read-only access and adding edit capabilities only when needed.
Follow the connection guide for setup and see the supported tools for what the Tape MCP Server can do.
Public beta note
This is an early version of the Tape MCP Server, and we’d love to improve it together with you.
Found something that doesn’t work as expected? Share it in Bugs & Issues.
Have an idea? Add it to Feature Requests.
Happy building.
