This issue: agents in every function, the data copies nobody tracks, receipts that cost nothing to fake, and the meter on every vendor's agent.
Buying an agent from every system of record recreates SaaS sprawl one layer up. How CIOs should choose, and which controls must stay central.
AI-generated fake receipts now beat visual review. How expense fraud changed, which detection still works, and the controls to use by company size.
Agents act through integrations. iPaaS and API vendors hold the connectors, credentials and audit trails, and MCP makes governing that layer worth more.
Every AI feature copies company data into prompts, logs, embeddings and memory. Where each copy lives, what vendors retain, and how to make deletion real.
AI is taking the bottleneck work out of the month-end close. The close becomes continuous with a short sign-off, and finance, audit and the board change.
Most tasks run better on one well-tooled agent. Multi-agent systems pay off only for parallel work, isolated context or independent checks.
Where AI helps in hiring and HR, where legal risk concentrates, what NYC, the EU, Colorado, Illinois and California require, and which controls hold up.
This issue: autonomy set by the action, approvals that get read, the index that owns the assistant, and spend limits on the card.
AI spend is moving from central IT to business lines. That ties cost to outcomes but breaks controls built for an IT-held purse. The model that works.
What AI can take over in accounts payable, step by step from capture to payment, where the fraud risk grows, and which controls must stay outside the model.
The first agents trusted with company money will buy for the enterprise. Control moves from approval chains to policy as code and card-level limits.
