Will we soon be negotiating with machines? – How AI is changing the game in business
Artificial intelligence analyses body language, recognises emotions in a fraction of a second and calculates the best argumentation strategies in real time. What does this mean for managers, buyers and sellers – and where are the opportunities and the dangers? A look at the future at the negotiating table.
The new negotiating table?
A grey conference room in Bentonville, Arkansas. Where Walmart negotiates prices, discounts and terms with its suppliers, there is now more than just a team of buyers sitting around the table. Invisible at the table is software that checks offers and calculates alternatives in a matter of seconds: payment terms, delivery times, raw material prices, transport costs, discounts, advertising subsidies. Whereas entire departments used to spend weeks poring over columns of figures, an algorithm now takes care of the routine business.
The Estonian start-up Pactum AI has developed a platform that automatically adjusts thousands of supplier contracts. Suppliers who respond to emails or fill out online forms often don’t even realise they are dealing with a machine. A machine that never gets tired, never seems irritated and never thinks of taking a spontaneous break.
For Walmart, this is a liberating move: negotiators can focus on the really difficult tactical issues in negotiations and save time for analysis and preparation. For the other side, it comes as a shock: suddenly, they are no longer negotiating with people, but with a system that is relentlessly efficient and checks and provides information in real time. Welcome to the new negotiating table.
From the lab to practice
What seemed like a gimmick from university laboratories just a few years ago is now everyday practice in global companies. Law firms such as A&O Shearman are reducing the negotiation time per contract by several hours with their ContractMatrix platform. The start-up HyperStart CLM combines automated redlining, risk reports and AI-supported decision-making support – and advertises time savings of up to 75 per cent in contract negotiations. Its customer list ranges from insurers to software companies.
German providers are also following suit. Statworx has developed an “AI negotiation assistant” that evaluates historical offers, material prices, inflation data and external market data to suggest negotiation strategies. There are still no major reference cases – but the direction is clear: AI systems are moving beyond pilot projects and into practical application.
Universities are also driving development and training forward. The MIT AI Negotiation Competition shows how pure AI systems fare at the negotiating table. There, programmes competed against each other to negotiate contracts. What was surprising was not so much that the systems were able to close deals, but how they did it: Dominant, demanding bots often failed, while the “emotional” systems, which appeared empathetic and signalled a willingness to cooperate, achieved the most successful results. Michelle Vaccaro, who provided scientific support for the project, sums it up: “Dominance does not pay off, even with machines. The bots that appeared empathetic and cooperative achieved the best results.”
This observation is more than an academic detail. It shows that even in digital negotiations, values such as trust, sympathy and fairness remain crucial – whether between humans or machines.
AI as a wingman
While bots act as independent negotiators in pilot projects, there are also successful approaches to using artificial intelligence as a discreet coach. This approach is being trained in practice at the Centre for Negotiation at Quadriga University Berlin with various AI tools to support negotiators.
The systems play to their strengths in the background – as invisible companions.
For example, a system called MindMeld is used to whisper tips to negotiators in real time during ongoing negotiations: alternative wording, ideas for compromises, hints on tone of voice. Those who accepted the advice performed measurably better. Samuel Dinnar, a lecturer at MIT, even sees this as an opportunity for people who have previously avoided negotiations out of fear: “AI can build bridges, especially for people who shy away from negotiations. It provides security, offers training opportunities and opens up negotiation skills to broader groups.”
Platforms such as Gong go even further, recording sales and negotiation talks and breaking them down into data points: Who spoke for how long? Which terms were used? When did the tone of voice fluctuate? The result is a coaching tool that shows salespeople where they come across as too dominant or where a pause would have had more impact.
Crystal works in an even more specific way and, from a European perspective, does not comply with data protection regulations. The US tool creates personality profiles based on public data – LinkedIn profiles, blogs, social media posts. The recommendation is then, for example: “Be direct, results-oriented, avoid small talk.” This is invaluable for salespeople because they can immediately adapt their language to the communication style of the person they are talking to. But in Europe, Crystal is highly controversial: according to the GDPR, the automatic creation of psychological profiles without consent is considered legally uncertain.
And then there is TAWNY, a Munich-based start-up that develops emotion AI. Using camera or sensor data, the AI recognises micro-expressions and stress patterns, i.e. those tiny signals that professional negotiators also train to recognise. This represents a paradigm shift in negotiation training: participants can see afterwards when they created pressure or lost trust. A live dashboard during virtual meetings is also conceivable. But this immediately raises questions: do we really want software to tell us in real time when our counterpart is getting nervous?
Balance between humans and machines
AI has long been more than a theoretical experiment. It is changing the way we negotiate – from preparation to real-time support at the table. But as promising as the technology is, it also raises new questions: What happens when both sides rely too heavily on algorithms? Will cultural nuances be levelled out when AI standardises communication patterns? And how can companies prevent a wrong prompt or hidden bias from ultimately overturning a million-pound deal?
The answer lies in balance. AI can sift through mountains of data, recognise patterns and run through scenarios. But creativity, moral judgement and the ability to build trust remain deeply human. Companies that want to successfully conduct negotiations in the future must therefore take a two-pronged approach: balancing data intelligence with human insight.
The new power at the table
The question of whether we will soon be negotiating with machines can already be answered today: we have been doing so for a long time. Pactum, Lexion, Tawny and other systems are already sitting at the table, often unnoticed but effective. They take care of routine tasks, sharpen analyses, provide clues – and thus change the playing field. Whoever has the better data gains power. Pactum and HyperStart show how algorithms can process standardised contracts at lightning speed. Companies that use such systems save time and resources and put pressure on their partners to keep up.
But power is not just about speed, it’s also about information. AI systems can recognise that suppliers have historically offered discounts more frequently in the fourth quarter, or that certain wording in contracts has regularly led to disputes. These analyses shift the balance: those who have AI behind them have an information advantage. This poses a challenge for small and medium-sized enterprises. They often do not have the resources to deploy complex AI systems themselves. This threatens to create a new asymmetry: corporations with AI power on one side, less well-equipped partners on the other.
In the end, it all boils down to one simple realisation: it is no longer a question of whether we will have AI in negotiations. It is already there, sometimes visible, sometimes invisible. The crucial question is: how do we use it?
The opportunities are enormous: faster processes, better preparation, more realistic training, less routine work. So are the risks: data protection issues, potential for manipulation, the loss of human empathy.
However, negotiations are and will remain a relationship business. Algorithms can recognise patterns and calculate options – but trust, creativity and moral responsibility remain with humans. The future of negotiation will therefore not be human versus machine, but human with machine.
And perhaps that is the real revolution: not the replacement of the negotiator by AI, but the expansion of their capabilities. The art will be to use technology as a tool – without losing sight of the human aspect.