AI agents / control towers
As agents become enterprise workers, companies need oversight, identity, policy, approvals, logs, runtime security, and workflow governance.
A separate decision dashboard for tracking whether new information supports, weakens, or changes an investing thesis before it reaches Portfolio OS.
As agents become enterprise workers, companies need oversight, identity, policy, approvals, logs, runtime security, and workflow governance.
Enterprises will avoid giving one model vendor full control of agent memory, evals, orchestration, and execution state.
Regulated digital dollars become financial rails for payments, settlement, treasury movement, remittances, and software-native money movement.
Project Arc points toward production-ready autonomous desktop agents: governed by ServiceNow AI Control Tower, connected to Action Fabric, and secured/supported by NVIDIA OpenShell and Enterprise AI Factory infrastructure. This strengthens the view that enterprise AI agents need both a workflow control layer and a secure runtime/compute layer.
Claude Managed Agents expanding into memory, outcomes, and multi-agent orchestration supports the view that enterprises will worry about vendor lock-in and control-plane ownership.
Committee vote on May 14 may advance stablecoin/crypto rules. Banks oppose the measure partly because stablecoin reward language could threaten deposits — useful signal that stablecoins are becoming a real financial-rail competitor.