If you’d rather not install the Intervals MCP integration but still want to use AI to work with your Intervals data, building your own tool is a practical alternative. This guide walks you through how to approach it.
The good news: you don’t need to be an experienced developer to build something useful. AI can do most of the heavy lifting. Your job is to know what you want to build and be willing to describe it clearly.
The Intervals API is well-documented — and AI can read it
The Intervals API is fully documented at myintervals.com/api, including a complete OpenAPI specification that describes every endpoint, resource, and parameter in a format that AI models understand well.
This means you can hand AI the API documentation and it will have everything it needs to start building. You don’t need to explain how Intervals works or what endpoints exist — AI can read the docs and figure that out. You just need to describe what you want your tool to do.
How to approach it
The most effective way to build with AI is to start with a clear description of the problem you’re trying to solve, not the technical solution. Don’t think about code, endpoints, or architecture. Think about what you wish you could do with your Intervals data that you can’t do easily today.
Some examples of the kinds of things people build:
- A weekly summary that pulls time logged across all projects and formats it the way your team actually wants to see it
- An alert that flags tasks that have gone quiet — no updates, no time logged — for longer than a set threshold
- A report that compares estimated hours to actual hours across active projects
- A tool that takes a description of work and logs it to the right project automatically
- A dashboard that shows each team member’s workload at a glance
These aren’t hypothetical. The Intervals API supports all of it. And AI can write the code that makes it work.
Starting the conversation with AI
When you’re ready to start building, open a conversation with AI and give it two things:
1. The API documentation. Share the OpenAPI spec or link AI to the relevant resource pages at myintervals.com/api. This gives AI a precise understanding of what the API can do and how to use it.
2. A description of what you want to build. Be specific about the outcome, not the implementation. For example:
“I want a tool that shows me every open task across all my projects, grouped by project, sorted by due date, and highlights anything that’s overdue.”
“I want something that I can run at the end of the week that totals up the hours logged by each person on my team and sends me a summary.”
“I want to be able to type a description of what I worked on and have it automatically log the time to the right project.”
AI will ask clarifying questions, propose an approach, and start building. You review what it produces, tell it what to change, and iterate from there.
What to keep in mind
Your API token is the key to your account — keep it off AI. When AI builds something for you, it will leave a placeholder where your token goes. You fill that in yourself when you run the tool locally. Never paste your actual API token into a chat window.
You don’t need to understand the code to steer the build. AI explains what it’s doing in plain English as it goes. If something doesn’t work the way you expected, describe the problem in plain English and AI will fix it. You’re the product manager; AI is the developer.
Start small and add from there. The best builds start with one specific thing done well, then grow. Resist the urge to describe everything you might ever want upfront. Get one thing working, then ask AI to add the next piece.
What you build is for your own use — not for the public. AI-assisted code is a powerful starting point, but it hasn’t been audited for security. Tools you build with AI are fine to run internally for yourself or your team, but you should not publish them, share them with people outside your organization, or expose them to the internet without first having a qualified developer review the code for security vulnerabilities. This is especially important for anything that handles authentication, accepts user input, or writes data back to your Intervals account.
API behavior differs from the web app in some ways. The most common difference you’ll notice is email notifications. By default, the Intervals API does not send email notifications when tasks, projects, or other records are created or updated — even though the web app does. If your tool needs to trigger notifications, the API supports this via special request headers. See the API documentation for details.
The API has rate limits. The Intervals API allows up to 100 requests per minute and 6,000 requests per day on most plans. For anything that runs repeatedly or processes large amounts of data, AI can help you build in caching and efficiency from the start — just mention it.