Somehow this is a contrarian opinion, but debuggers, code navigation, and autocomplete still matter. Developers are not being replaced.

Starting a codegen company

Kevin and I started Sweep two years ago.

In 2023 GitHub Copilot was the status quo.

We saw it and thought “this is not the final state of software development”.

"What do we do at our day jobs?" Almost all of our time was spent editing existing code.

We built the MVP of our GitHub agent in 4 days and applied to YC.

Going through YC

During our first office hour, we tried to clearly describe what we're building. "AI developer". Sweep will take GitHub issues or Jira tickets, and use LLMs to generate a solution. Then the developer can just review it.

We pushed forward - we open sourced it as well.

It was really fun! We built an awesome community. Hundreds of users per day, 5k github stars in a month, and over 1000 discord users.

Trying to deliver a coding agent

We pushed hard on getting agents to work. We spent our days fighting gpt-4-32k, then fighting claude opus 3, then fighting sonnet 3.5.

The biggest feedback we heard was "it's faster for me to do this myself. this doesn't work for anything non-trivial."

Even as we continued improving it felt like we couldn't make the leap.

We tried to take as much agency as we could.

Our mentality was "Ok the models suck, but we have so many ways around that".

We kept pushing and trying new models and techniques (future post) but throughout 2024, almost all of our users moved to Cursor. We were stuck.

How do we make agents work?

Of the major use cases of agents in coding, the ones that work well today are:

- Terminal agents (Claude Code) - IDE sidebars (Cursor, Windsurf)

Notice the pattern? These exist fully within the IDE or development environment.

There are three factors that play into each other.

1. Allowing agents to execute code is extremely complex or infeasible for large codebases. 2. This causes agentic coding loops to be bottlenecked on manual testing / user feedback. 3. As codebases grow in size, agents can't hold the entire codebase in context.

This means that for most codebases, agents still need supervision in 5 to 10 minute intervals.

Even with agents the natural place for developers to write code is still the IDE.

Pivot

We decided to pivot from our standalone agent. A lot of people were surprised.

I remember going to a developer meetup and telling a founder, "hey - we pivoted to an IDE coding assistant".

He gave me a doubtful look - "for VSCode?". There were too many VSCode forks at this time.

"No, for JetBrains!".

He asked, "doesn't Codeium have something already?"

I had already tried it. "It's way behind their VSCode fork."

We started Sweep with the goal of automating simple GitHub issues with AI. That mission became "help developers save time".

Building something people love using and saving them time is way more important to us than the way we get there.

We talked to hundreds of JetBrains developers, and they kept saying "the AI in JetBrains sucks."

Cursor had done a great job. Using AI in VSCode was seamless. But this wasn't anything magical. They had continuously iterated to find the best UX.

The previous tools that supported JetBrains, like Windsurf and Copilot, had just built a plugin to capture market share. It allowed them to position themselves as "one company for all of your AI needs."

They did not deliver on the JetBrains side.

Their users were constantly struggling with high CPU usage, missing features, and even outdated models.

So we decided to compete.

Over the last months, we've built an agent, trained an autocomplete model, and fixed A LOT of bugs. We're going to make JetBrains IDEs + Sweep AI the best way to develop.

If you're a JetBrains dev that wants to use AI without leaving your favorite editor, we'd love your feedback!

user:	gamifykaran
created:	November 24, 2023
karma:	2