AX is the only Experience that Matters - Ivan Burazin, Daytona

Key Concepts:

• Agent Experience (AX): How easily agents can access, understand, and operate within digital environments autonomously to achieve user-defined goals.
• Agent-Native Runtime: A computing environment specifically designed for agents, providing resources and tools they need.
• Autonomous Operation: The ability of an agent to complete tasks without human intervention.
• Declarative Image Builder: A system allowing agents to define and create custom Docker images on the fly.
• Daytona Volumes: Shared storage volumes accessible to multiple agent environments, facilitating data sharing.
• Parallel Execution: The ability of agents to run multiple tasks simultaneously for faster experimentation and problem-solving.

1. The Proliferation of Agents

• The speaker believes the number of agents will far exceed the number of humans, potentially reaching a scale of humans to the power of n.
• Evidence suggests a growing trend: 25% of YC startups report AI writing 95% of their code.
• 37% of the latest YC batch are building agents as their primary product, not just co-pilots or traditional SaaS.
• The speaker emphasizes that tools built for humans will become obsolete as agents become the primary users.

2. Defining Agent Experience (AX)

• Agent Experience (AX) is presented as the next evolution of User Experience (UX), Customer Experience (CX), and Developer Experience (DX).
• Sean from Netifi defines AX as "how easily can agents access, how easily can they understand and how easily can they operate within digital environments to achieve the goal that the user defined."
• The speaker highlights the importance of "autonomously" within this definition.

3. Implementing Agent Experience: Examples

• Seamless Authentication:
  • Problem: Agents struggle with authentication and should not be given passwords.
  • Solution: Arcade allows agents to fall back to user authentication when needed, or to use pre-stored credentials.
• Agent-Readable Documentation:
  • Stripe's practice of appending ".md" to documentation URLs for clean markdown files is cited as a good example.
  • The speaker urges the audience to implement the llm's.txt standard in their documentation.
• API-First Design:
  • The speaker argues that APIs are the most efficient way for agents to interact with tools.
  • Companies like Neon, Netifi, and Superbase are praised for providing comprehensive API access to their services.

4. The Importance of Autonomy

• The speaker stresses that true AX requires agents to operate autonomously, without constant human intervention.
• He challenges tool builders to consider what happens when there are no humans available to debug or click buttons.
• Swix challenged the speaker, suggesting that agent experience is just a wrapper around developer experience, which the speaker disputes if autonomy isn't considered.

5. Speaker's Background and Daytona

• The speaker has a long history in technology, including building data centers, creating the first browser-based IDE (pre-Repl.it), and leading developer experience at Infobip.
• Daytona is presented as a secure and elastic infrastructure specifically designed for running AI-generated code, essentially an agent-native runtime.
• Daytona aims to provide agents with a computing environment analogous to a laptop for a human.

6. Daytona's Approach to Agent Experience

• Daytona focused on speed, providing runtimes that spin up in 27 milliseconds.
• API-first design allows agents to programmatically manage machines.
• Headless tools (file explorer, Git clients, LSP terminal) are pre-loaded to avoid terminal parsing.

7. Advanced Features for Agent Autonomy

• Declarative Image Builder:
  • Allows agents to define and build custom Docker images on the fly, eliminating the need for human intervention or brittle agent-driven Docker builds.
• Daytona Volumes:
  • Enables agents to share large datasets across multiple environments without repeated uploads/downloads.
• Parallel Execution:
  • Allows agents to fork machines and run multiple experiments simultaneously, accelerating problem-solving.

8. The Future of Agent Experience

• The speaker admits that the full scope of AX is still unknown, as agent capabilities are rapidly evolving.
• He reiterates the importance of building tools that enable agents to perform tasks end-to-end without human assistance.
• The speaker concludes that agents will become the dominant users of technology, making agent experience the most important consideration for tool builders.

9. Key Quotes

• "The number of agents in the world I believe won't just match number of humans but they will basically be the number of humans to the power of n."
• "If you're building dev tools for humans you're basically building for the past."
• "How easily can agents access, how easily can they understand and how easily can they operate within digital environments to achieve the goal that the user defined." (Sean, Netifi's definition of AX)
• "If you haven't thought about or if you have a tool that at some point needs a human in the loop then you probably haven't solved or thought about how to solve this."
• "Agent experience is the only experience that matters."

10. Synthesis/Conclusion

The talk emphasizes the shift from human-centric to agent-centric software development. Agent Experience (AX) is presented as a critical consideration for building future-proof tools. The key takeaway is that developers should prioritize agent autonomy by providing seamless access, clear understanding, and efficient operation within digital environments. Daytona's features, such as the declarative image builder, Daytona volumes, and parallel execution, exemplify how to design infrastructure that empowers agents to perform complex tasks independently. The speaker urges the audience to rethink their development strategies and focus on enabling agents to achieve their goals without human intervention.