SpeakersAgendaPartnersWhy attendVenue
Locations ā–¼

Register your interest

Get your ticket

Partnership opportunities

Save ā‚¬100 on your pass

Call to action
Your text goes here. Insert your content, thoughts, or information in this space.
Button

Back to speakers

Yehya
Abouelnaga
Member of Technical Staff
Lovable
Yehya Abouelnaga is a Member of Technical Staff at Lovable, working across the Agent Builder, Model Training & Evaluation teams. Previously, he was a Staff Engineer at Personio and spent 7 years at Amazon, where he led global teams building large-scale ML systems for content moderation and launched Amazon's first customer-facing LLM feature. He holds an M.Sc. in Computer Science from the Technical University of Munich and a B.Sc. in Computer Science from The American University in Cairo. His research on driver distraction and posture classification has been cited over 900 times, including work presented at NeurIPS 2018. He has also been a founding engineer at two venture-backed startups: Minodes (acquired by TelefĆ³nica) and SchooLinks (raised an $80M Series B in Q4 2024).
Button
15 September 2026 10:00 - 10:30
Reining in chaos: Agents as infinite state machines
Most people can't code - but agentic tools promise to build software on their behalf. We put that promise to the test with a concrete case: building an MCP server inside an AI-generated app (Lovable), using an off-the-shelf coding agent (Codex, Claude Code) as the builder. What starts simple breaks down fast, and the reasons why point to deeper questions about agentic architecture - how much state an agent needs to track, when a general-purpose LLM isn't enough, and what "good enough" evaluation looks like for a task with no clear ground truth. We'll walk through what failed, what we changed, and what that implies for anyone building agent-driven tools for non-technical users.