In a groundbreaking development, the UK-based AI company Luminance demonstrated its proprietary large language model (LLM) by successfully negotiating a contract with another AI. The live demo took place on 7 November, marking the first time a contract was negotiated without any human intervention.
The AI model, named 'Autopilot', was trained on 150 million legal documents and was able to understand and analyze huge amounts of data in a matter of seconds, automating the review of routine contracts like non-disclosure agreements (NDAs).

Autopilot identified potential areas of disagreement in the contract, engaged in a negotiation process with another AI representing the opposing party, and made necessary changes to the clauses. The entire process was completed in a matter of minutes. The demonstration involved a simulated back-and-forth negotiation on an NDA between two imaginary legal entities. NDAs are commonly used in the legal profession and require time-consuming, prolonged scrutiny due to their strict confidentiality limits.
Autopilot is considerably more sophisticated than Lumi, Luminance's ChatGPT-like chatbot, designed to act like a legal 'co-pilot.'

While the AI handled the negotiation process, a live log showed suggested modifications to clauses, allowing lawyers to oversee the procedure and guarantee its accuracy. Human involvement was only needed at the end of the process to sign the NDA.

Why does it matter?

This development is a significant milestone in the progress of applying AI to the legal field. Founded in 2015, Luminance specializes in AI for legal professionals. By automating tasks usually performed by lawyers, such as negotiating contract terms, AI has the potential to simplify operations, freeing lawyers' precious time so they can focus on more complex and creative tasks. Industry-specific LLMs trained on private and specialized datasets are one of the latest evolutions, turning generalist models like OpenAI's ChatGPT into bespoke ones. We can expect to see more of these domain-specific AI models being developed, as they should deliver superior performance. Additionally, using a custom-made model provides focused expertise for firms looking to streamline processes with the benefit of data privacy.