Can artificial intelligence negotiate better than a human? Robots don’t get angry or upset; they don’t forget the details of what they’ve been told; they can process numerous pieces of information concurrently; they don’t care about how a person looks or sounds; and where many people might instinctively approach a negotiation as a zero-sum game, AI can be programmed to prioritise wins for both sides.
Large organisations with diverse procurement operations often have multiple neglected vendor deals that are set up to automatically renew when they expire. These are usually on unchanged terms and therefore vulnerable to new inflationary pressures.
Such oversights could be the result of poor negotiating to begin with, or simply a case of a company lacking the time or manpower to commit to poring over potentially thousands of different deals. A study by KPMG has suggested that up to 40% of the value lost in corporate deals can be attributed to inefficient contracting.
Improving long-tail contracts
Estonian software firm Pactum AI aims to square this circle. The company, which is now headquartered in Mountain View, California, but maintains an office presence in Estonia’s capital, Tallinn, has developed a chatbot technology which can negotiate with suppliers on a company’s behalf.
Pactum wants to help businesses “cut the cost of doing nothing”, quips its co-founder and chief product officer Kaspar Korjus. “The idea is to optimise cost saving, create revenue generation and expedite the negotiation process,” he says.
The firm operates across the retail, manufacturing, electronics and financial sectors, and can count the Fortune 500’s Walmart, Wesco and Maersk among its clients. Pactum’s AI-powered chatbot, which operates on a cloud-based interface that it invites suppliers to engage with via email, is designed to undertake contract negotiations with the long tail of vendors most large businesses use.
A company’s long-tail spend refers to the portion of its total spend that is made up of infrequent yet functional purchases. This might include facilities management, office equipment or services and technologies that a company uses to complete its day-to-day operations, rather than necessarily the products it sells to customers.
For example, Pactum’s first and arguably most high-profile client, the US retail giant Walmart, has used the technology to negotiate better terms with the supplier of its everyday store equipment, including shopping carts. In isolation, long-tail contracts might be inexpensive but, taken together, Korjus notes, they can add up.
Technology with roots in real-world experience
Pactum’s bot has been programmed and developed by a team that includes numerous human negotiation experts and psychologists.
Korjus was the managing director of e-Residency, Estonia’s government-backed digital identity system. Another of the company’s co-founders – and his brother – Kristjan Korjus, has a PhD in computer science and was head of data at Starship Technologies, a robotics firm.
The third co-founder, Martin Rand, was product manager at Skype and then at Monsanto acted as commercial lead for Europe for The Climate Corporation, an organisation which develops digitised solutions for farming.
Korjus rejects the idea that the bot is only as qualified or as capable as those who created it. “No,” he says, “that’s not right, because the AI learns new things continuously from each negotiation it takes on. So, it becomes more skilled and experienced with every deal, and it will use those lessons in the future. It will learn a lot more than just what we have taught it, and much faster too.”
AI as an expert deal negotiator
Pactum’s bot, Korjus explains, starts by collecting relevant data about the negotiation from the client and public sources, such as historical agreements, market trends and specific terms under consideration.
It engages with a supplier or vendor, following a “rule-based framework” set out by the client in advance. A client will tell Pactum’s human negotiation experts what it hopes to gain from a new contract. What terms does the company want and at what price? The bot needs to know what compromises the client is willing to make, as well as any red lines.
Having established what the client values and what scope it has for trade-offs, the bot will then ask the supplier or vendor a series of questions that prompt them to reveal their preferences and try to find a deal that works for both parties. In the back and forth that ensues, Korjus claims that the bot uses complex negotiating tactics and psychological techniques.
Pactum’s bot can negotiate thousands of deals at the same time, Korjus says, while clients can monitor their progress via The Negotiation Suite, an online dashboard.
Is AI worth the investment?
So, how much does all this cost? Pactum charges a variable six-figure annual sum, Korjus says, and moves to a seven-figure sum if rolled out globally. The company also charges the client a percentage of the gains from each new optimised contract.
Since it started working with Pactum in 2021, Walmart has said it has reached optimised deals with around 70% of the suppliers it has approached, with average savings of 3% on the contracts taken over by the bot.
“By allowing AI to handle your company’s negotiations,” Korjus says, “you can refocus your attention on strategic, high-impact decision-making. This means reallocating your team’s time and energy away from routine, time-consuming negotiations and towards tasks that require human creativity, strategic thinking and relationship-building.”
Using AI for efficient equality
It’s easy to see why Pactum’s technology may appeal to large corporations. But what about their smaller suppliers? Why would a cost-saving for a large company be good news for them?
“Our AI aims to identify Pareto-optimal outcomes,” counters Korjus. “These are situations where one party cannot benefit further without negatively affecting the other side. The most straightforward path towards achieving Pareto-optimal situations involves acknowledging and prioritising the value of the other party’s interests.”
Because the “best contracts are the ones with a long-term value”, Korjus explains, it makes sense for the larger corporations to keep their smaller suppliers happy. And, he points out, there are more ways to do this than simply giving them a higher fee.
He describes a scenario in which a large corporation uses a small plumbing company on one of its multiple sites. Korjus says that while the small company might accept a lower fee after negotiating with the bot for that one site, it could have plans to expand but no means as yet to make it happen. “This is the sort of question the bot will ask,” he says.
“Maybe the plumbing company will then also service the large corporation in other cities. In the long term, the plumbing company gets more business and can start up in new locations, so that is beneficial to them. It also creates an opportunity for more long-term trust in a partnership between the two parties.”
The potential of AI in HR
Pactum, Korjus says, is slowly moving away from simply the supply chain. The company is exploring ways in which the bot might assist with other negotiations, including those concerned with human resources.
It has already started to use the bot internally for negotiations about pay and hybrid working arrangements, but Korjus says there are still tweaks to be made before Pactum would feel comfortable rolling it out.
“Thinking about future applications, the deployment of this AI negotiator in other companies, even in the negotiation of aspects of people’s jobs, is a possibility,” he says. “However, there are potential risks to consider. The most notable is the possibility that people in similar roles might end up with widely divergent employment terms due to the parameters set before negotiations. While the AI negotiator is proficient at following predefined rules, the outcome may still vary based on individual preferences and priorities.”
Indeed, Korjus points out, workers who don’t give detailed enough responses to questions or effective enough prompts of their own to the AI might miss out compared to a colleague who does.
Nevertheless, Korjus is sure that these teething issues can be overcome. “The key to a successful AI negotiator in personal settings lies in striking a balance between customisation and fairness,” he says. “We need to ensure that individuals receive employment terms that align with their unique preferences and requirements while adhering to established guidelines for equity and fairness.”
Is AI the key to ending all disagreements?
Korjus makes a strong case for autopilot economics, particularly when it comes to managing sprawling supply chains. And Pactum can confidently claim that it has helped some large corporations save millions of dollars.
Yet there is a caveat to all this efficiency: it feels impersonal. Some of the most important deals people and companies make are often made over a drink, a meal or a round of golf. Pactum’s AI-powered chatbot might be hugely impressive, but it cannot replicate the feeling of shaking someone’s hand.
Indeed, could smaller suppliers feel slighted that a large business doesn’t think them worthy of human attention? How would that feeling tie in with Pactum’s mantra of long-term happiness and making sure both parties feel appreciated?
Korjus says feedback from suppliers has been overwhelmingly positive. They prefer negotiating with the AI, he suggests, because they can do so at their own pace, rather than in real time.
Ultimately, Pactum’s AI-powered chatbot has disrupted the negotiating landscape. That much is clear. But, at its six-figure price point, it seems fair to say that only a few companies will be able to consider using it.
