FAQs

What is Fine?

Fine is an AI coding agent that can carry out tasks around your codebase. It works in an isolated dev environment, independently and asynchronously, meaning developers can delegate tasks from anywhere.

How does Fine work?

Fine combines knowledge of your codebase (through integrations) with leading LLMs (such as o1 and Claude 3.5 Sonnet) to deliver agentic software development. At a basic level, when you give Fine a task, it collects the relevant context, creates a branch, writes the code and creates a PR for you to review and merge. Along the way, you can guide the AI, edit the code and give further instructions.

How is Fine different from tools like GitHub Copilot and Cursor?

Fine is an AI teammate. It lives outside of your IDE and instead of just suggesting code snippets based on what you’re typing, handles entire tasks that you assign it directly from Slack, Linear, GitHub or in the browser. We’re also completely mobile friendly, as we believe there’s no reason developers shouldn’t be able to get tasks done on the go, especially in the never-stopping startup world.

Who is Fine built for?

Fine is built for startup developers and tech teams seeking to boost output, speed up their software development process and polish their product. If you’ve got a small team of developers, a long backlog of issues to resolve and an even longer list of features to develop - Fine is for you.

What can I do with Fine?

With Fine, you can generate, test, and refine code, share live previews, automate repetitive tasks, and integrate with tools like GitHub and Linear to streamline your workflow.

While it works for variety of tasks and projects, Fine was designed and tested for a few specific use cases in mind:

• LLM and AI applications
• Backend applications and REST APIs
• Small UX/UI improvements

How long does it take to set up Fine?

Setting up Fine takes just a few minutes—simply connect it to your existing tools and you're ready to go.

Is Fine fully autonomous?

In some ways, sure. Think of Fine as another team member on your team. It has access to the codebase and knows how to perform lots of dev tasks.

Of course, Fine can’t push changes on its own - it creates a PR which you’ll need to merge. No need to worry about the AI running away, causing havoc.

But the good news is that Fine works asynchronously on its own, virtual machine. You delegate the task and when you’re ready come back to review. It’s far more independent than a pair-programming stye tool that just completes your sentences for you.

How can I get the most out of Fine?

Start by looking for small issues that will allow you and Fine to get to know each other.

Give plenty of information and context, tagging relevant files as necessary.

Review the implementation plan to make sure it’s what you’re looking for.

Make sure to Run the code to see that it works, before merging.

How can I trust Fine's generated code?

Nearly as much as you can trust a junior developer, yes! Thanks to the context awareness, Fine doesn’t make the basic issues that other tools do. And as you can Run the code in a virtual dev environment without leaving Fine, it’s easy to see if the code has worked in the way you intended.