83% of them already rate AI handling business relationships a top priority. The pull is easy to understand once you see where the hours go. 40% of these executives spend ten or more hours a week reviewing and approving contracts. That is twenty-one days a year, before a single negotiation even starts.

So the appetite is real. The harder question is whether the AI is any good at the actual negotiating, and MIT went and tested it. In an international competition the Initiative on the Digital Economy ran with 300 to 400 of the world's top negotiation experts, the winning bots were not the aggressive ones. Warmth beat pure killer instinct, and the strongest blend was warmth with dominance, the same mix that works for a skilled human across a table. The researchers also found a failure mode that only exists for machines: a counterpart can try to prompt-inject your bot mid-deal, a trick no human negotiation playbook has ever needed to cover.

Put those two findings together and a sensible starting point appears. An AI agent can carry the high-volume, lower-stakes contracts that eat your week, and it can hold a tone that stands up to a trained negotiator. The deals where the relationship or the dollar figure is too large to risk stay with you, at least until you have watched the agent earn that on the smaller ones. That is how an Agentic OS grows inside a business, one workflow at a time, which is also where an AI Audit would point you first.

Sources: Icertis C-Suite AI Survey 2025 (1,000+ executives); MIT Initiative on the Digital Economy, 2026.