Ex-GitHub CEO launches Entire to change how we build software
In this article, you’ll discover:
- Why current developer tools are hitting a breaking point in the age of AI.
- How Thomas Dohmke is building the world’s first assembly line for code.
- The secret to capturing agent context so you never lose the “why” behind a change.
- How to use Checkpoints to make your Git history searchable and machine-readable.
- Why Entire is raising $60 million to bridge the gap between humans and agentic intelligence.
The way we build software just hit a breaking point. If you have spent any time in a terminal lately, you have probably felt it. We are no longer just typing out lines of code, we are managing AI agents that generate files faster than we can actually read them.
The old tools we use, like Git, were built for humans talking to humans. They were never meant to handle a fleet of machines churning out thousands of lines in seconds. Thomas Dohmke, the former CEO of GitHub, noticed these cracks forming and decided it was time for a total platform shift.
Building an assembly line
Thomas compares our current situation to the early days of car making. Before the assembly line, cars were built by hand, one by one. Software is still stuck in that “craft” phase, even though we now have agentic intelligence doing the heavy lifting.
His new company, Entire, just came out of stealth with a $60 million seed round to fix this. The goal is simple: create a world where humans and agents can ship together without the system falling apart. It is about moving away from manual production and into a future built for machine speed.
The problem with “why”
We have all been there. You look at a piece of code from three days ago and ask, “Why did we do this?” When a human writes it, you might find a comment or a commit message. But when an AI agent writes it, that reasoning usually disappears the moment you close your terminal.
The decisions, the prompts, and the logic are lost forever. This forces the next agent to start from zero, often repeating the same mistakes. It is a waste of time and a waste of expensive API tokens.
Introducing Checkpoints
To solve this, Entire is shipping its first tool: an open source CLI. It introduces a new concept called Checkpoints. Think of it as a black box flight recorder for your code.
On every push, the CLI captures the full context of the agent session. It records the prompts, the files touched, and the reasoning. It then attaches this data directly to your Git history.
- Traceability: You can see exactly what the agent was thinking.
- Faster Reviews: You review the intent, not just a messy diff.
- Shared Memory: Agents can finally learn from the past instead of guessing.
A platform for everyone
What makes this exciting is that it is not a closed garden. Entire is designed to be model agnostic. Whether you use Claude Code, Gemini, or Cursor, the platform is built to be a universal layer for code collaboration.
The company is starting as a small, remote team of fifteen people who really care about developer experience. They are skipping the usual corporate fluff like OKRs and focusing on high tempo and quality.
If you are tired of losing track of what your AI tools are actually doing, you can get started with a simple curl command. It is a small step toward a much bigger vision: a future where the code tells its own full story.
