Summary
ING and FedEx described how they scale AI in high-trust, high-stakes environments where banking transactions and package delivery cannot tolerate uncontrolled experimentation. Both emphasized governance, guardrails, model orchestration, and production discipline, with examples including ING’s mortgage agents and Copilot-assisted engineering, plus FedEx’s customs automation and two-hour delivery predictions.
Key Topics
5 key topics from Daniele Tonella, Vishal Talwar and Harry McCracken at HumanX Amsterdam 2026. Thicker branches were mentioned more often in the talk.
Operating at Scale
Click a topic to open it
Key Takeaways
- Build governance before scaling AI into production systems.
- Use AI where it removes operational friction measurably.
- Keep autonomous agents away from live systems until trustworthy.
- Preselect models based on use case, economics, and data needs.
- Treat AI agents with human-level identity and access controls.
“We are both in the business of trust.”
Share
Up Next
RM
WatchNext in agenda
Trusting AI in the Wild
Robert Blumofe, Matt Rosoff
ai agentstrustworthy ai
GH
WatchAlso on enterprise ai
The New Luxury: AI is only as valuable as the hands that shape It
Gonzague De Pirey, Hanne Jesca Bax
ai adoptionluxury
JJU
WatchAlso on enterprise ai
Building Sovereign AI
Joelle Pineau, Jelle Prins, Umesh Sachdev, Rowland Manthorpe, Lisa Oder
sovereign aienterprise ai