Has software development entered the "black box" era? Former GitHub leader: In the future, no one will check the code written by AI

With the exponential evolution of artificial intelligence programming capabilities, the software development industry is standing at a critical point of complete transformation: human programmers may no longer directly review computer code, but instead completely delegate this heavy task to AI.

Former GitHub CEO Thomas Dohmke is betting on this future scenario with his newly founded company Entire, which aims to provide infrastructure for this "era where humans do not need to look at code." On Tuesday, it announced that it raised $60 million in seed funding, achieving a valuation of $300 million.

This vision is not unfounded; industry pioneers have already begun to practice it. Peter Steinberger, the creator of OpenClaw, recently stated that he no longer even reviews most of the AI-generated code used to build the AI agent. This trend is prompting a fundamental shift in development models from individual developers to large enterprises, and it has created an urgent demand for entirely new regulatory and monitoring tools.

In this context, Entire, founded by Thomas Dohmke, has emerged to reimagine the entire software development process. The company's core value proposition is that even if humans no longer review code line by line, businesses still need tools to understand the behavior paths of AI programming agents. This round of financing was led by Felicis, and notably, Dohmke's former employer, Microsoft, the owner of GitHub, also participated in the investment, demonstrating the tech giant's recognition of this transformation direction.

However, this "hands-off" model poses a severe challenge to corporate compliance. Dohmke warned that while developing applications without looking at code has become a technical option, no company can afford the risk of releasing "unreviewed code" from a safety and compliance perspective, as this could lead to serious legal liability issues. Entire aims to bridge the gap between the efficiency of AI autonomous programming and the necessary transparency required by enterprises.

Decoding the "Black Box": Monitoring Solutions Without Reading Code

The product logic of Entire is designed to address the "black box" anxiety brought about by AI programming. Dohmke pointed out that while individual developers like Steinberger can let AI handle the heavy coding work without review, large enterprises cannot afford such a completely "hands-off" strategy. Entire's goal is to develop a set of tools that allow people to understand the activities of AI programming agents without having to delve into the code itself.

As the first step in this strategy, Entire launched its first product on Tuesday—an open-source tool called Checkpoints. This tool can connect to developers' command line interfaces (the window where developers issue coding instructions to the AI agent) and record the operation logs of the AI agent in real-time. This includes not only the specific actions taken by the AI but also its gradual "thought" process.

This mechanism allows humans to obtain information about the operational status of AI-generated applications through the Checkpoints tool or to make repairs when potential issues arise. Currently, Entire supports Anthropic's Claude Code and Google's Gemini CLI (an interface that allows developers to issue coding instructions to the Gemini model) And plans to increase support for other programming agents in the future. Unlike developers like Anthropic, which focus on assisting the development of their proprietary agents, Checkpoints aims to monitor various AI agents from different manufacturers used by developers.

AgentOps (Agent Operations) Track Heats Up: Giants and Startups Compete

The establishment of Entire marks the intensifying competition in the emerging field of "AgentOps (Agent Operations, a comprehensive lifecycle management framework designed for autonomous AI agents)." This category includes startups and cloud services that observe or monitor AI agent behavior. Recently, several software providers have released similar products to help businesses more easily track the behavior trajectories of autonomous AI systems in their applications.

The market competition landscape is rapidly forming. Industry giants, including Microsoft and OpenAI, are actively promoting new agent monitoring products, trying to secure a place in this fast-growing market. Entire has chosen a commercially validated path of open-source software companies: first launching open-source tools and planning to add a cloud-hosted subscription service in the coming months.

From Microsoft Executive to Entrepreneur: Betting on Development Tool Transformation

Dohmke's entrepreneurial inspiration stems from a keen insight into industry trends. According to him, last summer, after witnessing the strong momentum of GitHub's AI coding tools, he conceived the idea of founding Entire. After discussing this idea with Microsoft CEO Satya Nadella and other executives, Dohmke decided to fully commit to product development and left Microsoft in August to establish the new company.

"The world of software development and development tools is about to undergo a dramatic change," Dohmke stated, "This is too exciting an opportunity to miss." This statement not only explains his motivation for leaving to start a business but also foreshadows a paradigm shift coming to the field of software engineering