Enterprise AI agents: the gap between prototype and production

Описание: Explore UiPath’s approach to agentic AI: https://uipath.ly/3O2fTaQ

Featuring UiPath AI Leaders:
Taqi Jaffri, VP, Product Management
Mircea Neagovici-Negoescu, SVP, Software Engineering

What does it take to build enterprise-grade AI for automation at scale?

In this conversation, UiPath AI leadership goes deep into how the UiPath AI team approaches model selection, fine-tuning, agents, document intelligence, and computer use—with a strong focus on real-world enterprise requirements.

You’ll hear how UiPath balances frontier models and open-source models, when and why fine-tuning is used, how agents continuously improve, and what it really takes to deploy AI reliably in cloud, private cloud, and on-prem environments.

Whether you’re an AI practitioner, automation leader, or enterprise architect, this session offers a rare inside look at how production-grade AI systems are actually built.

(This video was recorded in November 2025)

00:00 – Why enterprise automation needs new AI benchmarks
01:01 – The mission and structure of the UiPath AI team
01:50 – Frontier models vs open source: how UiPath chooses
03:10 – Fine-tuning strategies: shared models vs per-customer learning
04:20 – Why UiPath uses LoRA adapters at scale
05:45 – Deploying AI in cloud, private cloud, and on-prem
06:57 – How agents improve: prompts, in-context learning, and fine-tuning
09:45 – Computer use in production: what works today and what doesn’t
14:20 – Document intelligence: reading order, OCR, and GenAI tradeoffs
16:45 – Benchmarking enterprise AI
18:03 – The future of automation: coded agents and API-first design

🚀Join our community for more updates:
Academy: https://uipath.ly/462jPft
Blog: https://uipath.ly/3EwSRRC
LinkedIn: https://uipath.ly/44SelTOa
Facebook: https://uipath.ly/45A4naF
Forum: https://uipath.ly/3ReA7O3