AI engineer at Sona.
Writing agents in elixir at Sona to automate manual workflows.
I have worked for the last 2 years building out an LLM platform at Multiverse.
Founder of AdventOfSQL.com with over 3600+ members the site is run by AI Agents, Elixir, and LiveView.
I will explore the potential of Elixir in developing AI agents, challenging the predominant Python-centric approach.
The talk will explore why Elixir’s architectural and language features make it an ideal choice for AI agent development.
Key Discussion Points
- Foundational Architectural Considerations
- Analyzing the core components of AI agents: state management, concurrency, and decision-making processes
- The current best practices in architecting AI Agents
- Examining how Elixir’s BEAM VM, OTP, and functional programming paradigm fit the problem perfectly
- Technical Implementation Strategies
- Detailed exploration of GenServer for state management in AI agents
- Architectural patterns for building multi-agent systems using OTP
- Mechanisms for integrating external AI and machine learning services
- Leveraging Elixir’s concurrency model and fault-tolerance capabilities
- What the Elixir ecosystem needs to do to be the no-brainer choice in Agent development
Key Takeaways:
- Why AI Agents don’t require python packages
- How to build AI Agents in elixir
- Best practices in AI Agents
- What the Elixir ecosystem needs to excel in Agent development
Target Audience:
- Elixir developers interested in LLMs.
- Non-elixir devs curious about how Elixir can be an excellent choice for AI projects.