The Agentic SDLC: From Specs to Autonomous Merge

Join us for an evening exploring how engineering teams are moving beyond AI-assisted coding to autonomous workflows, from turning specs into production-ready code to building AI agents trusted to review and merge pull requests.

Agenda:

18:00 Gathering, mingling, food & drinks

18:30 Humans where they matter, AI where it’s better: Agentic SDLC at monday.com - Neyema Awaskar, monday.com

19:00 No Human Needed: building a merge agent developers trust - Omri Levy & Omer Militscher, Baz

Talk Abstracts:

Talk 1:
Humans where they matter, AI where it’s better: Agentic SDLC at monday.com
Neyema Awaskar, monday.com

A year ago, engineers wrote code, opened PRs, waited for review, and merged.
Tomorrow, they write a spec—and the rest happens automatically.
When generative AI hit the scene, monday.com engineers started shipping faster than ever, but we quickly hit a wall: writing code became easy, yet review queues kept growing. The bottleneck had shifted from writing code to everything that came after.
We realized that automating code generation was just bandaging a symptom. The cure required a complete paradigm shift: moving engineers from Builders to Directors.

In this talk, we’ll share what worked, what failed, how we built trust and how we measure impact. Finally, we’ll give you a practical starting point that you can implement tomorrow to take your first step toward full autonomy.

Talk 2: No Human Needed: building a merge agent developers trust
Omri Levy & Omer Militscher, Baz

Auto-merge sounds great until you ask the real question: would you actually trust an agent to merge your PR without looking? That's a question about trust - and trust is hard to earn.

At Baz we've spent the last few years building agents across the dev cycle: deep code review that map-reduces subagents over a diff, a spec reviewer comparing your tickets and Figma designs to the actual UI, and most recently a merge agent that decides whether a PR is ready to merge, no human needed.
In this talk we'll go under the hood of the patterns behind building agents that actually work - and specifically the merge agent: how it's triggered, how it reaches a verdict, how we test and benchmark it, and what's still left to figure out.

Speaker Bios:

Omri Levy is a squad manager and part of the founding team at Baz, building AI agents for the SDLC - from code review to production monitoring. Before Baz, he was at IBM Research, working on hyper-parameter optimization for RAG pipelines on the IBM watsonx platform.

Omer Militscher is a product manager at Baz, where he builds AI agents for the SDLC - from code review to production monitoring.
Before Baz, he was a Data Scientist at Salesforce, developing a data focused conversational agent, for Tableau Next Concierge team.