The Case of AI vs AI – How an AI Negotiated a Legal Contract with Another AI for the First time?

Updated on November 11 2023

For the first time since its inception, two AI negotiated a contract without any form of human involvement.

It has been 67 years since AI was first introduced, and it has evolved a lot since then. 

But who are the negotiators in this contract?

A UK based firm Luminance developed an AI model called “Autopilot.”

This AI model undertakes the use of proprietary large language model (LLM) which deeply analyzes and customizes contracts accordingly. 

It is a notable development, as the AI models that we have been dealing were limited to: 

  • Recognizing patterns and making predictions
  • Generating text, images, and other creative content
  • Automating tasks
  • Analyzing files we feed 

However, for the first time in the artificial intelligence industry, global population is witnessing decision-making approach and analysis by AI. 

Also Read: How did AI Evolve in its History of 67 Years? 

What is Luminance?

Luminance AI Negotiate legal contract

Established in 2016 by mathematicians, Luminance specializes in providing legal document analysis software designed to enhance the efficiency of legal professionals.

Utilizing an AI and machine learning-based platform, the company analyzes extensive, intricate, and fragmented legal data sets. This enables seamless task assignment and progress tracking for entire legal teams, leading to improved workflow management.

Luminance is backed by notable players of the market, including Invoke Capital—a venture capital fund established by British tech entrepreneur Mike Lynch—along with Talis Capital and Future Fifty.

Notably, Mike Lynch, co-founder of enterprise software firm Autonomy, has been a controversial figure facing extradition from the UK to the US due to fraud charges. Despite stepping down from Luminance’s board in 2022, Lynch maintains a significant backing role in the company.

Legal Contract Negotiated by AI with No Human Involvement 

Luminance, a British AI firm, built an AI system based on its proprietary large language model (LLM). This system autonomously analyzes and modifies contracts, aiming to streamline the cumbersome paperwork typically handled by lawyers on a daily basis.

Jaeger Glucina, Chief of Staff and Managing Director of Luminance, highlighted the goal of their new AI, Autopilot, to alleviate lawyers from routine negotiations. Autopilot is developed to handle day-to-day negotiations, allowing legal professionals to focus on creative aspects rather than being occupied with routine tasks.

In a conversation with CNBC, Glucina explained that Autopilot covers the entire negotiation process, from opening a contract in Word to negotiating terms and finalizing the agreement using DocuSign. The AI is not only legally trained, but also possesses an understanding of the business involved.

Autopilot surpasses Lumi, Luminance’s ChatGPT-like chatbot, by operating independently of human intervention. While human oversight is still possible, the software logs all changes made by the AI throughout the process.

In a live demonstration at Luminance’s London offices, the AI-driven negotiation showcased remarkable speed. Clauses were analyzed, modifications were implemented, and the contract was finalized within minutes.

Legal Autopilot involves two lawyers on either side of the agreement—Luminance’s general counsel and the general counsel for one of Luminance’s clients, ProSapient. However, the driving force behind contract analysis, content scrutiny, and recommendations are entirely AI-driven.

Glucina highlighted the significance of Autopilot in expediting non-disclosure agreement (NDA) negotiations, a traditionally time-consuming process in the legal profession. With legal teams spending approximately 80% of their time on routine document review and negotiation, Autopilot’s ability to highlight contentious clauses, make suitable changes, and adhere to companies’ negotiation preferences offers substantial efficiency gains.

Luminance’s Autopilot, tailored specifically to the legal industry, distinguishes itself from more general-purpose AI platforms like ChatGPT and Dall-E. Industry experts, including Peel Hunt, anticipate the superior performance of domain-specific AI models over general-purpose ones.

While Luminance didn’t disclose the cost of its software, the company offers annual subscription plans to clients, including major entities like Koch Industries and Hitachi Vantara, as well as various consultancies and law firms.

How did the AI Negotiate the Legal Contract?

Autopilot’s efficacy as an artificial intelligence negotiating entity can be ascribed to several pivotal factors:

How did AI Negotiate with each others successfully?
How did AI Negotiate with each others successfully?
  • Extensive Dataset Utilization: Autopilot underwent training on an extensive dataset comprising human negotiations, encompassing transcripts, videos, and diverse data forms. This comprehensive training facilitated Autopilot’s acquisition of discernment regarding patterns and strategies employed by adept negotiators.
  • Reinforcement Learning Mechanism: Employing reinforcement learning, Autopilot continuously enhances its negotiating proficiency. The mechanism involves rewarding Autopilot for successful negotiations and penalizing it for unsuccessful ones. This iterative process enables Autopilot to discern effective strategies tailored to varying negotiating scenarios.
  • Adaptive Capability: Autopilot exhibits a remarkable capacity for adaptability, allowing it to dynamically tailor negotiating strategies to different opponents and contexts. This adaptability stems from Autopilot’s continuous learning and iterative improvement mechanisms.

Furthermore, Autopilot employs a distinctive strategy of generating multiple offers concurrently. This approach expeditiously explores diverse negotiation options, enabling Autopilot to identify and present the most optimal offer promptly. Leveraging its profound understanding of human psychology, Autopilot strategically incorporates knowledge of human biases into its negotiation tactics. For instance, Autopilot adeptly structures offers to align with human cognitive biases, increasing the likelihood of their acceptance.


AI has come a long way with necessary developments and changes. It is yet to be seen how it aids the technological market which is ever evolving and what lies ahead of the human-AI relationship. This article talks about how AI closed a legal matter without human intervention, which imposes a serious challenge; will AI replace humans, or we will adapt as we evolve?

Featured Tools







Related Articles