Thomas Dohmke

Thomas Dohmke is co-founder and chief executive of Entire, a developer-platform company building infrastructure for collaboration between people and AI coding agents. As GitHub’s chief executive from 2021 to 2025, he oversaw the expansion of GitHub Copilot, helping establish AI-assisted programming as a mainstream software-development practice.

Raised in Germany, Dohmke studied computer engineering at Technische Universität Berlin and earned a doctorate in mechanical engineering from the University of Glasgow. He worked at DaimlerChrysler and Bosch before co-founding HockeyApp, a mobile-developer-tools company acquired by Microsoft in 2014. At GitHub, which he joined in 2018, he helped introduce free private repositories and became chief product officer before succeeding Nat Friedman as CEO in 2021.

His tenure encompassed Copilot’s evolution from editor autocomplete into conversational assistance, GitHub Copilot Workspace, and GitHub Models. Early experiments with OpenAI’s Codex convinced him that generated code became useful when embedded directly into familiar workflows, where developers could evaluate suggestions without repeatedly leaving their editors. Copilot Workspace extended that approach from an issue or plain-language task through repository analysis, editable implementation plans, proposed changes, and pull requests, keeping developers involved throughout the process.

Dohmke left GitHub in August 2025 and launched Entire in February 2026 with $60 million in seed financing. His priorities include:

• Earned autonomy for coding agents: Agents should gain independence through demonstrated reliability, with actions remaining transparent, reviewable, and proportionate to the stakes. His writing on agent autonomy applies that principle to code review, security fixes, and other engineering tasks requiring human trust.
• Agent context as a software artifact: Entire’s Checkpoints connects agent sessions with Git history, preserving prompts, decisions, and execution context alongside code changes. Its open-source command-line tool makes machine-generated work easier to review, troubleshoot, and understand.
• Infrastructure for agent-scale development: Entire is developing distributed Git infrastructure, repository mirroring, and context-aware review for environments where coding agents generate changes continuously. Dohmke has also outlined native repositories and intent-driven development, emphasizing infrastructure sovereignty and software workflows designed for agents.
• Human-centered developer reinvention: Dohmke argues that AI changes programming without eliminating the need for judgment, curiosity, and experimentation. His account of developers adapting to AI describes an evolving craft, while his AI Engineer World’s Fair conversation connects natural-language tools with access to unfamiliar codebases, institutional knowledge, and programming across languages.

Dohmke continues building personal software, including hadge, which exports iOS Health workout data into a Git repository, and serves on Deutsche Telekom’s supervisory board.