AI Agent Strategy
summary
SAP describes a unified agent framework called Juul that spans HCM, CX and supply chain with four interaction patterns: informational, navigational, transactional, analytical. A live example shows a dispute case agent invoked over Microsoft Teams that analyzes a billing dispute, proposes fixes, opens the S/4 invoice, drafts the email, cancels and recreates invoices and logs actions in ESM. Emphasis is on system-triggered agents and decision support, democratization via Dual Studio, benchmarking multiple LMs and about 10ish agents in beta. Knowledge graphs and a one‑domain model are positioned as the next enabler.
AI-written summary · about
chapters (9)
- 0:00AI agent strategy intro
- 1:04SAP’s Juul agent framework
- 3:29Dispute case agent demo
- 7:40Agents as decision tools
- 9:47Democratizing agent dev with Dual Studio
- 13:20Autonomy vs determinism in agents
- 15:49MCP limitations for enterprise agents
- 17:56Future of SAP agents and knowledge graphs
- 20:43Q&A
related talks
42mAI Agents in Action: Transforming Enterprise Processes
26mAI agents 101
11mAI Agents for CX
45mAn Agent Walks Into a Bar... and Meets Another Agent: Multi-Agent Systems in Action
40mAgentic AI: The Next Evolution of Automation
46mAI Agents for the Rest of Us
49mAI Agents: Collaboration in Real-Life Applications
33mAgent Experience