Is AI negotiating with AI coming sooner than we think?
I witnessed something new in a Nibble negotiation recently. A shopper was chatting with our AI bot, asking for a discount when they paused and said: “No no, let me go chat to another AI to help me. It’s called ChatGPT, I will be back.” The customer literally went to recruit a second AI to assist them against the first AI. This real conversation snippet is a sign of what’s coming: AI negotiating with AI – not in some distant sci-fi future, but soon, and in some places, already happening.
A Glimpse of AI-vs-AI Negotiation Today
That shopper’s strategy of pitting one AI against another might have been lighthearted, but it highlights a trend. We’re starting to see early forms of AI-to-AI negotiation. For example, enterprise use of AI negotiators is growing fast – one recent survey found 36% of procurement teams already use AI for parts of their supplier negotiations (8 Stats That Show Trends For AI In Procurement In 2024). And it’s not just experiments: global companies like Maersk and Walmart have deployed AI agents to negotiate supplier contracts at scale. In these cases, an AI on the buying side may be negotiating deals with a human or an AI-assisted seller. The logical next step? Both sides using automated agents. Gartner analysts even predict an explosion of “custo-bots” – customer-side bots negotiating deals – in B2B transactions. But how soon might AI-on AI negotiation become common?
Consider that these negotiation AIs are improving rapidly, and adoption is accelerating. Salesforce’s futurists project that 2025 will be the year “agents talk to agents,” negotiating and transacting without human intervention (Future of AI Agents 2025 - Salesforce). It might start with simple negotiations – your AI assistant bargaining with my AI sales bot over a bulk discount – and then move up to more complex contract negotiations. One industry veteran even predicts we’ll see multiple AI negotiation unicorns (billion-dollar startups) by 2030, underscoring how quickly this field is evolving. In other words, AI-agents brokering deals with each other may become a business-as-usual scenario within a few years, not decades.
Why Let Bots Bargain: Efficiency, Scale, and Even Fairness
The promise of AI-to-AI negotiation is efficiency. Machines don’t need breaks or eight hours of sleep. Two well-designed bots could theoretically hammer out an agreement in seconds or minutes. In fact, companies using AI negotiators today are already seeing negotiations finish in minutes rather than weeks.
Imagine sending out 100 contract renewals on Friday at 6 PM and finding a stack of completed, mutually-acceptable agreements on Monday morning.
AI negotiators also scale effortlessly – a single agent can handle thousands of simultaneous mini-negotiations (something no human team could do). This kind of always-on, high-speed bargaining could slash cycle times and liberate humans from tedious back-and-forth on routine deals.
There’s also a case to be made that AIs, in some situations, might find a fairer middle ground faster.
Without ego or emotions, two negotiating algorithms could quickly zero in on the zone of possible agreement. Each can be programmed to stick to pre-approved limits (so they won’t agree to something unsustainable), and they can transparently log every offer and counter. In theory, this transparency and consistency might build trust in the outcomes. Early data from autonomous negotiation pilots is encouraging – for instance, organisations report supplier satisfaction improved by 20% after using AI negotiators, likely because the process became faster and more objective.
When neither party feels bulldozed into submission and instead feels the outcome is data-driven, it could lead to more win–win deals.
New Challenges: Complexity, Control and Unexpected Surprises
If AI-on-AI negotiation is the next frontier, it comes with its own set of worries. One big consideration is control and oversight. When two autonomous agents are wheeling and dealing, how do we ensure they don’t go rogue or make bizarre decisions? This isn’t just theoretical – there have been cases where dueling algorithms produced absurd outcomes. A famous example: two pricing bots on Amazon once entered a feedback loop and drove the price of a book up to $23.6 million before humans intervened (Amazon Algorithm Price War Leads to $23.6-Million-Dollar Book Listing). Worth remembering as Amazon just launched its shopping agent.
There’s a risk of a negotiation arms race: whose bot has the better algorithm or more data? If not carefully governed, AI-to-AI negotiations could exacerbate power imbalances – imagine a small supplier’s simple bot up against a mega-retailer’s sophisticated bot. We may need “rules of engagement” for AI negotiators to prevent exploitative tactics and ensure outcomes stay within ethical and legal bounds (after all, what an AI agrees to can be as binding as a human contract – who remembers the Air Canada chatbot that promised a customer a full refund?)
The Path Ahead
All signs indicate that fully autonomous AI-to-AI negotiations are coming soon, and they could fundamentally change how businesses interact. In the near future, your procurement bot might routinely negotiate with a vendor’s sales bot while you sleep – and they’ll do it faster and maybe even better than you could.
But the one we are all waiting for? The bot which can renew all my subscriptions for me like Sky TV or Virgin Media broadband and prevent the hours lost on hold to call centres and on live-chat.
The implications are huge: deals struck at unprecedented speed, richer data informing every offer, and a lot of mundane negotiation taken off human plates. But it also means we’ll have to carefully design these agents with guardrails, empathy, and ethics in mind, so that “let the bots handle it” truly yields fair and trustworthy results for both sides. Be considerate about who you buy your bot from.
Find out more from Nibble's experience negotiating 100,000 times a month here.
Interested in Nibble?
