September 17, 2025
An Unfiltered Look at Agentic AI Engineering
From exciting demos to real enterprise agents
AI agents are everywhere—but most conversations stop at flashy demos.
The hard truth? Building agents that actually work in the enterprise takes more than a foundation model.
Most conversations about AI agents are all about the flashy demos—what they could do
But let’s be honest: the real story is in the not-so-glamorous stuff, the building blocks that make an agent actually work in the enterprise.
That’s where the magic—and the hard truths—live.
In this session, Uber AI Solutions will break down the five essential building blocks every enterprise needs
We’ll share lessons learned from building at Uber-scale—where agents don’t just run in a lab, they face millions of real-world edge cases every day.
You’ll learn why skipping the unglamorous steps leads to brand failures, and how Uber’s testing frameworks and data capabilities give enterprises confidence to deploy agents safely, accurately, and fast.
- Fine-tuned LLMs that truly reflect your company’s tone, values, and context
- Clean, reliable data sources that ensure accurate reasoning
- Strict access rights for safe, multi-tool environments
- End-to-end workflow scripts that define how agents reason, call APIs, and respond
- Rigorous agent testing frameworks—because if you can’t test it, you can’t trust it
Featured articles
Uber AI Solutions for Agentic AI
Industry solutions
Industries
Guides