Your AI Contract Says No. Can the Agent Still Do It?

Hey! I'd love to hear your thoughts, send me a voice note.

What happens when your AI contract requires human approval — but the system can still allow the agent to act?

In Episode 3 of Procurement at AI Speed, I look at the gap between contractual authority and technical enforcement.

A rule written into a contract or policy only goes so far if the technology can still execute the action.

Using a simple procurement scenario, I explore three questions worth bringing into the next sourcing and contract discussion:

• What is the agent actually allowed to do? • What technically enforces those limits? • And who owns those controls after the contract is signed?

I also look at NVIDIA’s recent announcement of technology designed to enforce boundaries outside the AI agent itself — not as a product recommendation, but as a signal of where the market is heading.

The practical takeaway:

Don’t just ask whether an AI control exists. Ask to see what happens when the agent tries to cross it.

Because telling an agent what it can do is one thing. Technically stopping it from crossing that line is another.

The $100,000 approval threshold used in this episode is an illustrative scenario.

Procurement at AI Speed is a short-form series exploring what AI changes commercially for procurement — and where human judgement still matters.

Hosted by Zina. By KaiZina.

Explore KaiZina https://kaizina.com.au

Connect on LinkedIn

Zina Amir Moafi https://www.linkedin.com/in/zinaamirmoafi/

KaiZina https://www.linkedin.com/company/kaizina/

Sources & further reading

NVIDIA — Open Agent Safety Platform, 28 September 2026 https://nvidianews.nvidia.com/news/open-agent-safety-platform

Australian Signals Directorate — Careful adoption of agentic AI services https://www.cyber.gov.au/business-government/secure-design/artificial-intelligence/careful-adoption-of-agentic-ai-services