The use of artificial intelligence by buyers and suppliers to negotiate price and terms is in the early stages of acceptance.

Many customers today still aren’t ready to fully finalize deals with AI involved. But in the next two to three years, for simple and non-critical procurement categories, we’ll see a full handoff of negotiations to AI on both sides — buyers first, and suppliers a bit later. After that will come automation of more complex negotiations and systemic changes.

AI-driven negotiations will no longer have to “die” on weekends and holidays. We’ve already seen active negotiations aided by AI agents on Christmas Eve and New Year’s Eve, with suppliers sending clarifications, lowering prices and improving terms late into the night.

Over time, AI will can move negotiations into a 24/7 mode, without pauses dictated by the calendar. And when AI agents begin operating on both sides, the time from sending a request for proposal to closing a deal will shrink dramatically, from weeks or months to a matter of hours.

In international procurement, the effect will be even stronger. AI agents remove delays tied to time zones as well as national holidays.

The mass adoption of AI in negotiations will inevitably hit those who make money from information asymmetry, in the form of financial and informational arbitrage. If negotiations and decision-making are performed by systems that see the market more broadly and react more quickly than humans, intermediary margin starts to disappear.

What’s more, if market participants use comparable large language models, then demand-and-supply coordination won’t happen through competition among brokers, but through the underlying architecture of AI systems.

We’re already seeing early signs of these effects in specific scenarios — for example, when errors and biases in trading algorithms on exchanges lead to anomalies and “jittery” market behavior. In negotiations and procurement, this path is only beginning to form, but the direction is obvious.

As negotiations are delegated to AI, and automation advances in robotics, an entire ecosystem of companies built on information arbitrage will start to break down. Most likely, future architecture will remove the majority of intermediaries, and the market will consist instead of AI agents that negotiate, optimize terms and coordinate processes, and operators in the physical world who perform real actions, such as logistics, warehousing, production and service.

The operational layer will remain, but management and coordination will become much more “machine-driven.” At the top level, an agent will appear that receives requests from people and other agents, and below it will be a set of specialized agents managing supply chains and execution.

Full integration of this stack is a hard problem, so don’t expect a complete solution in the next 10 years. But individual elements are already being automated at scale — ports, warehouses and logistics have been moving in this direction for a long time.

Automation of supply and demand will affect not only business, but also political mechanisms. Today, many market processes are still regulated manually, through decisions by regulators and specific individuals. The more processes are optimized through AI, logistics, distribution systems and robotics, the less room remains for manual control and manipulation.

On the one hand, this increases the efficiency of production and distribution. On the other, it reduces the ability to apply political pressure to these processes, because they become more transparent and more resource-efficient.

Regulation of AI systems in distribution and coordination, as well as robotic systems, will most likely emerge. But we’re still far from mature, well-designed regulatory mechanisms. The technological shift is moving faster than institutional change.

In the coming years, expect to see procurement negotiations in simple, non-critical categories being largely delegated to AI. Buyers and sellers will begin treating negotiation AI the way that financial markets treat algorithmic trading: as an always-on default that rewards speed, pattern recognition and disciplined execution. As this happens, spreads will tighten, execution will accelerate, and the value of sitting “in the middle” will shrink — meaning most intermediary margin will disappear.

In practical terms, the ecosystem restructures vertically, with interface agents at the top; negotiation and coordination agents in the middle, and real-world operators —factories, ports, warehouses, shipping, last-mile and service crews — at the bottom, all tightly orchestrated by software. This efficiency gain will also disrupt established influences built on access, timing and information.

The use of AI in negotiations compresses transaction time and gradually eliminates large classes of arbitrage, moving closer to the concept of the perfect market: one in which many buyers and sellers have full information; firms can freely enter and exit, and no participant can influence the market price, which is set purely by supply and demand.

Vitaliy Goncharuk, an American entrepreneur specializing in autonomous navigation and AI, is chief executive officer of A19Lab.com.