GitHub CEO Thomas Dohmke Warns Startups: AI Can’t Replace Human Developers
Paris, June 14, 2025 – Speaking at the prestigious VivaTech conference held at Station F in Paris, GitHub CEO Thomas Dohmke struck an important note on artificial intelligence’s evolving role in software development. While acknowledging the remarkable impact of AI coding assistants like GitHub Copilot, he issued a clear caution: startups cannot scale—or attract serious investment—without building real technical depth.
Dohmke argued that although AI tools have significantly lowered the barrier to entry, enabling non-technical founders to launch companies, they fall short in delivering the kind of robust, complex systems that investors prize. He stated:
“A startup built predominantly with AI coding tools doesn’t have much value to investors … a non-technical founder will find it difficult to build a startup at scale without developers”.
The Core Message: Blend AI with Expertise
At Station F, Dohmke emphasized two key points:
- AI lowers entry thresholds by helping non-developers prototype quickly.
- Human expertise remains irreplaceable: Scaling, security, maintenance, and innovation require experienced developers who understand system architecture and code intricacies.
This underscores a broader industry consensus: while AI tools enhance productivity, they are still best utilized under human guidance, especially in mission-critical or high-scale software.
Broader Vision: AI Agents in Developer Workflows
Dohmke has also shared a roadmap for GitHub Copilot’s evolution. According to his recent post on LinkedIn, the new Copilot Agent Mode now supports multi-call prompting (MCP), improved chat tools, and richer source control integration within Visual Studio Code.
He envisions a future where AI coding agents augment developers, handling routine tasks while enabling engineers to concentrate on complex design, architecture, and problem-solving. Importantly, he insists coding literacy remains foundational: developers must still understand underlying logic and security implications.
Why This Matters
| Trend | Implication |
|---|---|
| AI in coding | Widens who can prototype, but don’t eliminate expertise-driven development |
| Startup funding | Investors seek strong technical foundations beyond quick demos |
| Developer roles | Shift toward overseeing AI agents, securing systems, and managing complexity |
Bottom Line
Thomas Dohmke’s message is both a celebration and a caution: while GitHub continues to advance AI coding tools, true long-term value still lies in deep engineering talent. AI can accelerate development—we need not define it entirely.