Autonomous negotiation at ocean scale: How Maersk uses AI to close freight contracts
A.P. Moller — Maersk runs one of the largest container shipping and integrated logistics businesses in the world. Every week, the company negotiates thousands of short-term commercial agreements: spot freight rates, inland haulage, warehousing, demurrage arrangements, tail-spend procurement. Many of these deals are small enough that dedicated commercial lawyers and procurement specialists can't justify the time — but large enough in aggregate to move the P&L.
Maersk's legal and procurement functions started with a provocative question: what if the negotiation itself could be handled by AI, with humans defining only the guardrails?
Working with Estonian AI company Pactum, Maersk's teams mapped the space of contracts that shared three characteristics: high volume, broadly standardised commercial structure, and a clear set of levers (price, volume commitment, payment terms, liability caps) that could be negotiated within a pre-approved envelope.
These were precisely the contracts that humans were either rushing through or leaving on the table. The in-house legal team identified the clause-level risk boundaries — indemnities, liability, governing law, force majeure — that had to remain inside Maersk's standard positions no matter what. Everything else became terrain the AI could negotiate on.
.jpg)
Maersk deployed Pactum's autonomous negotiation agent to run chat-based negotiations with suppliers and counterparties at scale. The AI opens the conversation, proposes terms inside the pre-configured envelope, responds to counter-offers, and closes — or escalates to a human when a supplier pushes outside the guardrails.
Legal's role shifted from reviewing individual deals to configuring the envelope:
The approach flipped the traditional model. Instead of lawyers reviewing contracts after commercial terms were agreed, they now set the rules up front — and the AI handled the conversation.
The measurable results were significant. Pactum's deployments with Maersk achieved approximately 15% savings on negotiated rates — a number large enough to justify the program on cost alone, before any productivity benefit.
Just as importantly, supplier satisfaction actually improved in many cases, with reports citing gains of over 20% in supplier experience scores. Counterparties valued the speed, consistency and 24/7 availability of the AI negotiator more than they minded negotiating with software.
For Maersk's legal function, the shift is strategic. The team has publicly signalled an ambition to let AI handle up to 80% of non-strategic procurement negotiations, freeing lawyers and commercial leads to focus on the high-stakes agreements — strategic vendor contracts, vessel-sharing agreements, port concessions — where human judgement genuinely moves the needle.
