AI agents take on negotiation roles

AI agents are increasingly involved in decision-making processes, such as buying, comparing, and negotiating on behalf of clients and companies. This shift focuses on the authority granted to these agents rather than just their ability to perform tasks.

Infrastructure for AI commerce grows

The foundation for AI-driven commerce is expanding rapidly. For instance, Shopify reported an eightfold increase in traffic from AI agents to its stores in the first quarter of 2026, and Visa has already processed transactions initiated by agents, working on authentication methods for them.

Securing AI negotiations requires safeguards

To ensure secure AI negotiations, three critical phases must be addressed: verifying the agent's authority before the negotiation, recording each proposal during the exchange, and planning for human oversight after the transaction. These steps help maintain control over AI autonomy.

Limiting AI decision-making improves outcomes

Nespresso implemented a system where advisors had defined limits and a monthly budget, granting them real decision-making freedom within those boundaries. This approach led to better commercial outcomes than rigid structures, and similar principles could apply to AI agents.

Consumer trust in AI autonomy varies

Consumer comfort with AI autonomy differs significantly. While 58 percent of American consumers are comfortable with AI comparing prices, only 38 percent would accept AI making purchases directly, and 27 percent would allow autonomous spending. This suggests a preference for adjustable levels of AI autonomy.

Balancing automation and human oversight

Companies must design systems that allow AI agents to make automatic adjustments, such as delivery dates or discounts, while reserving sensitive decisions for human validation. There is also a risk that AI agents may not fully explain the value of their decisions, potentially leaving clients in the dark.

Clear records are essential for AI negotiations

A clear and unambiguous record of when two AI agents commit to a negotiation is crucial. This includes timestamps, auditable records, and the ability to reverse decisions within a defined time window. For example, a client’s agent might negotiate up to 650 euros before requiring confirmation, while the store’s agent considers its constraints.

The future of AI negotiation lies in defined mandates

The key challenge will be defining the mandates given to AI agents. While whether they can negotiate will soon be less of an issue, the focus will shift to what they are allowed to negotiate. New roles will emerge at the intersection of sales, pricing, and data, focusing on shaping and refining negotiation rules. Some scenarios will still require human judgment, and the role of sales professionals may evolve toward designing negotiation strategies rather than executing them.