Agentic AI in the Wild: Open Source Agents in Production

AI and Open Source Builders Meetup, New York CityAI agents are moving into production, and more and more of them are built on open source. The question is no longer whether to use them, but what it takes to make them reliable once the demo is over. This meetup brings together builders, maintainers, and practitioners integrating AI agents into real codebases, CI/CD pipelines, internal developer platforms, cloud native infrastructure, and observability stacks. The evening focuses on the conversations that rarely fit into conference talks: what breaks at scale, how open source tools and standards shape agentic systems, where failure modes hide, and how teams build systems they can trust. Under the theme Open Source in Production, we will explore agentic code review, agent loops in CI/CD, developer platforms, eBPF-powered observability and security, and the role open source communities play in how production AI gets built. No vendor pitches. Just practical lessons from systems running in production. Agenda 5:00 PM Doors Open, Drinks and Networking5:30 PM Welcome and Opening Remarks5:40 PM [Speaker TBD], Qodo Dev Community6:05 PM Break, Food and Networking6:25 PM Panel: Agentic AI in the Wild7:15 PM Ryan Hamilton, Senior Developer and Community Lead, CircleCI AI Lab7:40 PM Closing Remarks7:45 PM Networking until 8:00 PM Speakers and Sessions [Speaker TBD], Qodo[Talk title TBD][Talk abstract TBD] Ryan Hamilton, Senior Developer and Community Lead, CircleCI AI LabOne Take: Engineering Agent Loops That Ship 100% Green PRs on the First PushMost teams optimizing AI-assisted development are focused on the wrong variable: prompting harder, adding rework, dumping larger failure logs into context. The real question isn't how to fix red PRs faster. It's how to stop producing them.Ryan shares findings from three connected Loop Lab experiments, each using the same controlled Snake game benchmark to isolate one variable at a time. Each one reveals something unexpected about where the real leverage lives in an agent loop, and what it means for cutting the cost of agentic development.Together, these experiments map a clear path from "rework is inevitable" to "rework is reserved for the unknowable." The result is a practical harness design that bakes a low Merge Efficiency Ratio (MER, the number of validation cycles it takes to get a change onto main) in from the start, and a new way to think about where validation belongs in the agent loop. Panel: Agentic AI in the WildWhat does it take to move AI agents from proof of concept into systems that operate reliably, and what role does open source play?Practitioners from across the stack will discuss context management, governance, open standards, observability, security, and what changes when agentic systems scale.Expect concrete lessons, not talking points. ModeratorDana Fine, Open Source and Developer Relations Manager, QodoDana leads open source programs and community initiatives at Qodo. She runs the GitHub User Group, CNCF local and GenAI communities, and the Bond AI meetup series. PanelistsBaruch Sadogursky, PortBaruch is a longtime developer advocate and a well-known speaker at developer conferences around the world. He co-authored Liquid Software and DevOps Tools for Java Developers, and focuses on developer productivity, DevOps, and how AI is changing the way software gets built and shipped. [Speaker TBD], Claude Community[Bio TBD] [Speaker TBD], NVIDIA[Bio TBD] Community PartnersAWS User Group NYC, CNCF NYC, Frontend NYC, WeMakeDevs, Bond AI, Qodo Dev Community Join the CommunityQodo Dev Community on Discord: [Discord invite link]Follow us on Meetup: [Meetup group link]