Anders Hejlsberg on the shift from AI assistant to AI agent

By GitHub

Categories: Product, Tools

Summary

The shift from AI assistants to AI agents will dramatically change development tools, with AI now doing the work and developers supervising rather than coding directly. This shift is enabled by connecting language services to MCP and giving AI the ability to ask semantic and refactoring questions.

Key Takeaways

1. AI assistants are being replaced by AI agents that can do the work themselves, with developers supervising rather than coding directly.
2. Connecting language services to MCP and giving AI the ability to ask semantic and refactoring questions is enabling this shift to AI agents.
3. AI agents can now perform the equivalent of what developers would do in an IDE, but in an LLM or agent-based way.
4. This shift to AI agents will dramatically change development tools, though the long-term balance between AI and human developers is still unclear.
5. AI agents no longer need an IDE in the same way that human developers do, as they can leverage language services and MCP to perform tasks.
6. The shift from AI assistants to AI agents marks a significant change in the role of AI in the development process, moving from a support tool to an autonomous agent.

Topics

• AI Agents
• MCP Language Services
• AI-Powered Development Tools
• AI Assistants to AI Agents Transition
• Semantic and Refactoring AI Capabilities

Transcript Excerpt

There's this whole switch that has happened in AI where where AI started out being the assistant, but you were still in control sitting in the ID and it was just aiding you in typing faster and completing stuff faster, right? But now it's switching around. The AI is doing the work, right? And and you're supervising what it's doing and and it doesn't necessarily need an IDE in the same way that we need idees. it still needs the services and that's why like all of this MCP stuff is becoming intere...