Episode Details

TNews includes Mark announcing his new Sagents library for building AI agents with human-in-the-loop oversight and real-time debugging capabilities, José Valim's blog post on why Elixir is the best language for AI based on recent completion rate studies, LiveDebugger v0.6.0 with custom events and enhanced exception handling, an analysis of Elixir developer salaries showing US senior remote positions at ~$163k median, Peter Ullrich's new MCP server implementation for Elixir, a revamp of Lua for Elixir with improved tooling, Elixir Hub's new events section for finding conferences and meetups, and a discussion on whether AI advancements like Opus 4.6 are eliminating the SDK advantage of more popular languages, and more!e

Show Notes online - http://podcast.thinkingelixir.com/292

Elixir Community News

• https://paraxial.io/ – Paraxial.io is sponsoring today's show! Sign up for a free trial of Paraxial.io today and mention Thinking Elixir when you schedule a demo for a special offer.
• https://github.com/sagents-ai – Github organization home to the Sagents projects
• https://github.com/sagents-ai/sagents – Mark announces Sagents - "Sage Agents" library combining wisdom with LLM-based agents, featuring HITL, SubAgents, GenServer architecture, Phoenix.Presence integration, middleware system, and more
• https://github.com/sagents-ai/sagents_live_debugger – Phoenix LiveView dashboard for debugging and monitoring Sagents agents in real-time with visibility into agent execution, message history, tool calls, and events
• https://github.com/sagents-ai/agents_demo – Demonstration Phoenix LiveView application showcasing Sagents's multi-conversation agent architecture with isolated AI agents and real-time updates
• https://x.com/josevalim/status/2019422467446698493 – José Valim announces his new blog post on why Elixir is the best language for AI
• https://dashbit.co/blog/why-elixir-best-language-for-ai – José Valim's blog post explaining why Elixir is best for AI, focusing on immutability, documentation, stability, and tooling for coding agents
• https://tidewave.ai/blog/the-future-of-coding-agents-is-vertical-integration – Study showing Elixir had the highest completion rate across models among 20 different languages
• https://www.reddit.com/r/elixir/comments/1qwqg1p/livedebugger_v060_custom_events_enhanced/ – Reddit discussion about LiveDebugger v0.6.0 release
• https://github.com/software-mansion/live-debugger – LiveDebugger v0.6.0 with custom events, enhanced exceptions, and temporary assigns support