Tomer Brook
Tomer Brook is a Senior Engineering Manager at monday.com, leading the team behind its Agentic Developer Platform and Atlas, one of its agentic developers.
When Agents Own Delivery: The Architecture of AI Teammates in Production
AI Coding Summit NYC
AI Coding Summit NYCUpcoming
When Agents Own Delivery: The Architecture of AI Teammates in Production
What if your AI agent wasn't just a tool, but a teammate? Not a chatbot you query, but an engineer who picks up tickets, opens PRs, ships to production, and is accountable for what it merges. We learned this the hard way, and the architecture required to make that work is fundamentally different from anything in the current playbook.
Most teams treat AI agents as assistants, but at monday.com, AI agents are real teammates and colleagues: each has a name, a role, and a team. They join engineering squads, get assigned tasks, reply in Slack threads, review PRs on GitHub, ship features to production, and monitor what they shipped.
This session explores the architecture behind that shift - from a task assignment to a shipped feature. We'll cover the event-driven platform design, how agents retain context and memory across multi-day tasks, the operational patterns that make them reliable at scale, and what changes when an agent becomes accountable for delivery in production.
To anchor the above, I'll introduce one of them- Atlas. He shipped code that passed locally and broke in production, and forgot what he'd worked on between sessions; we'll walk through the failures that taught us to make him trustworthy. Atlas is one of 1,000+ agents now running in production. Together, they've merged 2,000+ PRs into a 10-year-old codebase with 400+ services and 2M+ users, 80% of them zero-touch, at a 1% revert rate.
The audience takeaways will be: (1) the event-driven architecture pattern for agent-as-teammate systems, (2) how to design persistent memory across multi-day autonomous tasks, (3) the guardrails and eval loops that keep agents reliable at scale - and a clear view of where a human still belongs in the loop.
Most teams treat AI agents as assistants, but at monday.com, AI agents are real teammates and colleagues: each has a name, a role, and a team. They join engineering squads, get assigned tasks, reply in Slack threads, review PRs on GitHub, ship features to production, and monitor what they shipped.
This session explores the architecture behind that shift - from a task assignment to a shipped feature. We'll cover the event-driven platform design, how agents retain context and memory across multi-day tasks, the operational patterns that make them reliable at scale, and what changes when an agent becomes accountable for delivery in production.
To anchor the above, I'll introduce one of them- Atlas. He shipped code that passed locally and broke in production, and forgot what he'd worked on between sessions; we'll walk through the failures that taught us to make him trustworthy. Atlas is one of 1,000+ agents now running in production. Together, they've merged 2,000+ PRs into a 10-year-old codebase with 400+ services and 2M+ users, 80% of them zero-touch, at a 1% revert rate.
The audience takeaways will be: (1) the event-driven architecture pattern for agent-as-teammate systems, (2) how to design persistent memory across multi-day autonomous tasks, (3) the guardrails and eval loops that keep agents reliable at scale - and a clear view of where a human still belongs in the loop.