World's Fair 2025
Ship Agents that Ship: A Hands-On Workshop - Kyle Penfound, Jeremy Adams, Dagger
Kyle Penfound , Jeremy Adams , Dagger
Overview
This talk demonstrates how to build and deploy AI agents that can actively contribute to software development workflows using Dagger. It emphasizes creating agents with specific tools and environments, enabling them to perform tasks like code editing and testing within a sandboxed, containerized system. The core idea is to integrate AI capabilities into existing software engineering pipelines, providing guardrails and structure to manage the output of generative AI.
Who should watch
- AI Engineers looking to integrate agents into development workflows.
- Product Managers and Builders interested in leveraging AI for prototyping and shipping software.
- Developers seeking to automate code generation, testing, and deployment processes.
- Anyone exploring how to add structure and reliability to AI-driven development.
Key takeaways
- Dagger acts as a workflow engine and container runtime, allowing software engineering workflows to be portable and consistent across environments.
- AI models can be integrated as components within Dagger, treating them as another tool or building block alongside containers and code.
- Agents can be given specific, refined environments and tools (e.g., file editing, testing capabilities) to perform targeted tasks, rather than a broad, unconstrained set of functions.
- Dagger's declarative approach allows for the creation of typed inputs and outputs for agent environments, ensuring predictable interactions.
- Agents can be developed as Dagger functions, enabling them to leverage existing Dagger modules and the Daggerverse ecosystem.
- The process involves defining an agent's environment, prompt, and tools, then executing it within the Dagger framework, which can be visualized using tools like Dagger Cloud.
- Agents can be deployed to run within CI/CD pipelines, such as GitHub Actions, by integrating Dagger as a workflow step.
- Dagger provides a unified cache and a consistent API across different language SDKs, facilitating cross-language interoperability for modules and agents.
Notable quotes
*Dagger is for software engineering workflows and environments.*
*We're building the robot body. And the brain is going to come in just a second here, which could be any LLM kind of a brain in a jar.*
*The main thing is like you want the right amount of tools for that agent to solve its task.*
Unofficial community note. Prefer the recording for nuance.