Your next contract negotiation might be with a machine

Kaspar Korjus
Pactum AI
Published in
12 min readJun 8, 2020

Pactum’s artificial intelligence has surpassed human negotiation skills. Here’s why that’s a good thing — even if you’re the one negotiating with it.

At our early days in Pactum discussing our own biases of doing trade-offs in negotiations (co-founders Martin Rand on the left and Kristjan Korjus on the right)

Negotiations can be nerve-racking.

So much of our life quality is determined by the outcome of them — from where we live to how we work, as well as how much free time and money we have to engage in yet more negotiations. For companies, it’s even more fundamental. Good negotiations build successful businesses with quality products and services, motivated employees, and satisfied customers.

Good negotiating is an art, but I want to show you how we are now changing it to a science.

We’ve built an AI that can complete complex contract negotiations in just a few clicks and provide a better deal for all sides. It’s already being used by a wide range of companies and if you’d like to get more value from your suppliers too then I hope to make a case for using AI to have more effective business negotiations.

Like most people, I’ve never had any kind of training in how to negotiate. But I had to learn fast when tasked with building an organisation for the first time.

My background

Five years ago, I negotiated myself a job with the Estonian government to launch a new idea called e-Residency. We envisioned a future in which all countries would eventually compete to offer the best services online to entrepreneurs globally.

During my four years as Managing Director of the e-Residency programme, we negotiated with every part of the state, as well as Parliament, to turn this idea into reality and make it work. It wasn’t enough to have the support of the startup community in our capital. We needed to speak to the granny growing onions on her rural home farm or the low paid labourer concerned about globalisation so that they could see how they benefit from supporting this idea.

There are now more than 60,000 e-residents for Estonia with more than 10,000 new Estonian companies, generating more than €1 billion in revenue, and making a significant contribution to our country’s economy, including by paying more than €35 million directly in taxes. And that vision of the future is already reality. Several other countries have launched or are preparing to launch their own versions of e-Residency — from Dubai to Lithuania — while Estonia is already developing e-Residency 2.0. This time, negotiations aren’t necessary. It already has the rare consensus support of all political parties.

Along the way, we came up with a second big idea. What if Estonia became the first country to launch its own crypto tokens so that global entrepreneurs could invest directly into our country’s development? We called this idea Estcoin and it generated intense discussion around the world. I’ll be honest though. My attempt to negotiate Estcoin into existence failed to get off the ground. In addition to insurmountable opposition here in Estonia, even the Head of the European Central Bank publicly slapped down the idea.

I’ve now returned to business and, for the third time in my life, I’m incredibly excited about a new idea that I think can radically change our world. And, once again, I’m working with an incredible team that is passionate about turning this idea into reality. We’re already negotiating with some of the world’s largest and smallest companies, some of which are already using it.

Will it succeed like e-Residency or fail like Estcoin?

Well, this time, our success depends not just on how well we negotiate, but whether we can change the very nature of how companies negotiate.

Our idea is Pactum. It’s an Artificial Intelligence software designed to negotiate contracts on behalf of companies. While most software is designed to help people do business, Pactum is actually doing business for them.

In fact, we believe in the near future that machines will take on almost all operational work while humans can concentrate on strategy and creativity.

It’s now been 12 months since the initial idea with my two co-founders, Martin Rand and Kristjan Korjus, 9 months since the launch of the company and then 8 months since world class investors backed our idea, including former Skypers Jaan Tallinn and Taavet Hinrikus. You can read our initial blog post here.

It’s no longer just an idea though now. The software is already being used by paying customers to handle their supplier negotiations. We are particularly proud of how the world’s largest company, Walmart, is finding it valuable.

The art of AI negotiation

Here’s the thing about negotiating. It may be an incredibly important skill, but most of us are remarkably bad at it.

It’s not just a lack of training though. Negotiating goes against our very nature because pursuing our own immediate needs is woven into our DNA as animal instinct.

In negotiations, we need to look beyond that yet we struggle to weigh up all the complex considerations involved because there is too much data to consider in too little time. Instead, we let our own egos and biases shape the outcome to our own detriment.

Humans just aren’t rational in negotiations. An aversion to loss can overpower our ability to appreciate far greater gains. For example, companies are frequently reluctant to make simple concessions they can afford, like longer payment in terms, in return for deals that could significantly boost their revenue. A ridiculous opening offer can anchor a certain value and lead to us agreeing an outcome that’s still too close to that ridiculous offer. There are more than 60 other negotiating mistakes like these based on human psychology.

People tend to think of it as a zero sum game in which there must be a winner and a loser. But this is also a mistake. It’s more often the opposite: every accepted trade-off makes the deal better for both parties and should be considered a win-win, not a win-lose.

But there is a science to good negotiations.

The ideal negotiation results in something called a Pareto efficient deal. That’s an agreement in which both sides can’t get a better deal without harming the interests of the other side.

We initially wanted Pactum to learn by observing professional negotiators, yet we quickly discovered that even experienced negotiators waste too many opportunities for both sides and rarely achieve Pareto efficient deals at the end.

You might think that large companies at least have sophisticated processes and enough power to get what they want. But a KPMG analysis found that the average commercial deal could be improved by up to 40%. That’s because 80% of commercial agreements are high volume and low value. That’s called the ‘long tail’ — or ‘unmanaged tail spend’ because it’s so difficult for employees to keep on top of them and consistently negotiate anywhere near optimal deals on a consistent basis.

How Pactum is changing negotiations

Companies need to focus on their strategy, but instead of leaving suppliers unmanaged, Pactum is able to manage the long tail for them by negotiating the optimal win-win scenario in a vast number of complex contracts simultaneously.

Live negotiations between Pactum and the vendor

Our AI produces better contracts through automation than if a human even had the time to negotiate them all instead. We’ve discovered that every deal Pactum has negotiated has increased the value of the deal on average by 11%-27% for our clients.

And when I say ‘better contracts’, I mean for both sides. The simplest way to increase value for the client is to increase value for the supplier too. It’s always a win-win outcome we’re aiming for. More suppliers are accepting more deals negotiated with Pactum than they previously did with human negotiators, which is the clearest evidence that they also like the outcome.

Our AI saves time and human resources. So far, we’ve found that 96% of our clients’ negotiations are fully automatic using Pactum without any human involvement needed at all.

This isn’t automation that you switch on and forget about though. That time saved can be used instead for developing your strategy, which is informed by the unprecedented level of data about these contracts that Pactum can deliver.

Not only that though, but 76 to 82% of the suppliers we surveyed afterwards actually preferred dealing with our chatbot style interface. They said it’s because they were able to reach agreements faster with less hassle and, most importantly, with better terms for them. Perhaps most surprisingly, some even mentioned that the AI was polite and friendly. Human to human negotiations can get pretty tense, afterall.

The best part is that Pactum is constantly learning from its own negotiations and improving.

How Pactum works

Our artificial intelligence software works best in repeatable, rules-driven, high volume commercial negotiations. It can’t negotiate Brexit (yet!), but it can help companies handle the majority of their suppliers.

Automation has already proven itself successful at handling processes of this nature, such as by automatically paying invoices. This is known as Robotic Process Automation (RPA). However, this is the first time that it has been applied to negotiations, which can then shape and improve the actual processes too — whether they are automated or not. We’re calling this Robotic Process Enhancement or RPE.

It works because Pactum has learnt how to talk about business, as well as the limits of what can be offered on a vast number of factors and how to value the trade-offs between them. Humans can only focus on a small number of trade-offs, like the price and basic payment terms. Beyond that though, it becomes a serious mathematical challenge when considering factors like exclusivity, payment periods, quantities, demand, and margins. Only a system like Pactum can handle that properly.

Pactum triggers the conversation with ideal suppliers, as set out in the company’s strategy, then speaks to them through a chat-like interface to help them reach a deal that works better for both sides. The company using Pactum decides the overall strategy, priorities and limits. Then the negotiation flow is determined by data and various triggers that keeps the supplier updated.

The interface matches the client’s brand, including the URL, logo, and color scheme. That link is valid for a limited number of days, but can be reopened at any time from any device to continue negotiations where they were left off. The supplier can write text or numbers, as well as choose from multiple choice answers where necessary.

A deal is only reached if both sides find a better agreement than if they walked away and it’s no longer possible to improve it without harming one side’s interests. Here’s what that looks like on a graph.

Pareto optimal line to which all negotiations should lead

The X axis represents the value of the deal to the company that has decided to use Pactum. The Y axis represents the value to their supplier. Each dot represents a different contract being negotiated. The red curve represents the Pareto optimal line to which all negotiations should lead. It’s unbelievable to think that according to KPMG 80% of contracts negotiated by humans never reach that red line, which wastes value for both sides.

The AI is designed to find the tradeoffs that enable both sides to move closer to the red line as efficiently as possible. Every company has different needs so the AI will listen to what the supplier values and then figure out what can be acceptably offered to help the client and supplier get as much value as possible.

The Pareto optimal deal is so much a part of Pactum’s DNA that even our logo is a representation of it.

The logo of Pactum representing Pareto optimal deal

At the end of this process, the contract is automatically created and can then be digitally signed by the supplier before the Enterprise Resource Planning systems are automatically updated.

Our AI is learning fast

It’s never before been possible to conduct A/B testing on something as open and complex as negotiations, which is why scientific research on the subject is incomplete and contradictory. Yet Pactum is repeatedly learning by conducting negotiations in a relatively controlled environment with insight into thousands of data points. This is producing both better results for each client as Pactum understands their business better, but also improves the overall performance of Pactum for everyone.

Take a look at the results from 100 recent negotiations using three different iterations of the Pactum AI.

First, the preference of suppliers for using Pactum over a human negotiator rose from 50 to 74%. This is because the bot simply became better at communicating and was able to provide a more hassle-free experience. In addition, the percentage of suppliers who agreed to the final deal increased from 60 to 81%. This shows how well Pactum was able to listen to what the client wanted while moving the final agreement to the Pareto optimal line.

Most importantly though, the average value of the deal to our client using Pactum increased by 11% in added revenue. Through data science and machine learning, Pactum is able to constantly improve its strategies, tactics, and offerings.

The future of negotiations

Pactum is already at work for the largest companies of the world. Our team of developers and data scientists are also constantly working to improve both the AI and the user experience.

Today, the software has proven its ability to improve a company’s bottom line by optimising its contract negotiations on a large scale. Where do we go from here?

Let’s imagine you are the Chief Merchandising Officer for a new retailer.

It’s 2021. You start your day at work by opening Pactum’s dashboard, which by that point will provide enhanced oversight to the user. You see that the AI is conducting hundreds of negotiations simultaneously with both existing and potentially new suppliers. You can also look through the new negotiations that have been concluded in just the past 24 hours alone. The deals have already been digitally signed, your systems have been updated and supplies are already being loaded into vans and moved across the country to your warehouses as per the terms of the deals. The suppliers are happy. You can see the number of successful deals is up and so too is their net promoter score.

It’s 2023. In addition to even more oversight, Pactum has more in-built tools for you to choose the company’s strategic direction directly through the dashboard. By this point, your business has been growing steadily in large part due to Pactum. The automation of vast numbers of contracts allowed you to quickly keep adding products to your store with favourable terms. The oversight of these contracts combined with the time saved enabled you to focus on what products were selling the best and likely to sell better in future so you could then instruct Pactum which types of suppliers to focus on triggering negotiations with. In addition, the growth of your business has changed your needs and priorities so you can instruct Pactum to relax some of your previous negotiating red lines. You’ve got plenty of money in the bank, for example, so let your AI go softer on payment terms to get even better deals.

It’s 2025. Pactum is now giving you deeper insight into your strategic options. Your company has grown exponentially. But just as the economy is recovering from the pandemic, there is yet another black swan event that causes enormous disruption globally. This time though, you are prepared with Pactum. The AI looks at the impact of this disruption on your suppliers and consumers then explains to you a range of options you can take based on their probable impact on the bottom line. You instruct Pactum to diversify the supply chain and add more local suppliers to improve the resilience of the supply chain to the situation. You also refocus the target products towards those most likely to be in demand based on how you think consumer habits will change through this crisis.

It’s 2030 and you’ve been using Pactum for a decade in which your company has grown through the good times and the bad. Except, Pactum is no longer an interface used by your company. It’s your company’s first AI employee that sits in the boardroom and helps directly advise on the company’s decision making. By the time the meeting ends and the way forward is agreed, your AI employee is already implementing thousands of new actions simultaneously.

Start this journey with us

Pactum is still a new idea, but we are working with companies in various industries from retail to logistics, from the music industry to internet marketplaces. Now is the best time to come onboard with us and find out how our AI can work for you.

To find out more, write to me directly at kaspar@pactum.com. It won’t take long until a robot takes this job from me too, of course.

I’ll give you a 10 minute demo of Pactum then we can figure out how we can apply it to your business needs. By automating negotiations, you may find new sources of value that so far you have only dreamed off.

--

--