How AI Is Reshaping Supplier Negotiations

Harvard Business Review

3 days ago

AI is starting to play a major role in how companies negotiate with suppliers. These advancements are driven by the need for greater speed, scalability, and strategic agility due to increasingly complex supply chains and disruptions caused by external factors such as weather and trade wars. Once a cost-saving tool for automating low-value, repetitive negotiations, AI is now being used to make key decisions in procurement.
However, adopting AI isn't straightforward. Companies must navigate hurdles such as data quality, regulations, and matching the right tools to the right situation. Drawing from our more than six years of research on human-AI collaboration and the use of AI to conduct autonomous supplier negotiations, we outline in this article how companies can unlock the full value of these advances. Our research includes studies of major companies at the forefront of applying AI in supply-chain management in a variety of industries, including retailing, pharmaceuticals, consumer packaged goods, logistics, and IT.
What AI Can Do in Supplier Negotiations
AI tools are getting better at handling the nuance of real-world negotiations. By 2027, Gartner predicts, half of all companies will use AI-powered tools to help negotiate supplier contracts. To keep abreast of the pack, companies need to begin building the following capabilities now.
The Right Use Cases
Most companies start small, using AI for simple, repeatable purchases—like packaging or raw materials. But the value goes beyond that. When products or services don't vary much from supplier to supplier, AI can compare offers and focus on optimal terms like price, delivery time, and reliability. Consider how Walmart began experimenting with AI negotiation tools in certain product categories and quickly expanded its use. The shift wasn't just about cost savings; it was also about speed and supply chain agility.
Real-Time Market Awareness
AI tools can track supply and demand, pricing trends, and competitor behavior as it happens. This is a tremendous benefit for categories affected by price swings or frequent renegotiation. Pactum, a supplier of AI for automating supplier negotiations has demonstrated that using chatbots for negotiations at a massive scale improves working capital, increases supply chain resilience, and cuts costs. Henkel applied this capability to products impacted by volatile prices, and Maersk used it for freight services within existing supplier and customer agreements or to automatically secure quotes when none was available.
Operational Contextual Intelligence
Companies can now combine internal data (like operations, budgets, and supplier scorecards) with external forces (like regulatory shifts, currency fluctuations, and geopolitical risks) to shape negotiations dynamically. This integration enables AI to tailor negotiations as contexts evolve—for example, by allowing companies to detect tariff announcements in real time and adjust sourcing, pricing, or logistics. Similarly, IDEXX Laboratories used AI to determine which of its 70+ global suppliers were vulnerable to Russian sanctions, allowing it to proactively adjust contracts.
Smart Trade-Offs
Currently, generative AI is also evolving to assess trade-offs more precisely, balancing cost, sustainability, delivery time, and financial risk. Instead of focusing only on price, AI can recommend the best supplier partnerships based on company goals. For example, L'Oréal's procurement strategy uses AI to negotiate sourcing deals that balance cost and sustainability for key cosmetic ingredients. A health company we studied uses a digital advisor to help negotiators balance market dynamics, pricing models, negotiation strategies, predictors of negotiation rates, and procurement terms to meet strategic targets across categories.
The Shift Toward Autonomous Negotiation
Companies aren't jumping overnight from manual processes to having an AI system close deals on their behalf. Instead, they're progressing step by step, moving from AI-assisted to semi-autonomous to fully autonomous negotiations.
The Assisted Stage
AI tools act as copilots but do not replace human decision-making. While some autonomous actions may occur, they don't extend to external parties.
For example, Luminance's Legal-GradeAI system can generate automatic alerts. A multinational utilities company uses that feature to identify commercial risks and early-contract-renewal opportunities that include discounts. However, the system won't automatically send a renewal contract. Instead, it will flag the event internally, draft the contract, and may even prepare the renewal email.
Similarly, San Francisco-based Regrello, which offers an AI operating system for manufacturing and supply chain management, uses AI agents to draft terms, flag risks, compare clauses, gather approvals, and integrate contractual terms with other operational systems. Its AI system can also simulate scenarios—for example, modeling a 15% tariff on key materials to anticipate cost impacts and prepare for price renegotiations—but leave it to humans to make the final negotiations.
Semi-Autonomous Stage
Such systems can accept pre-approved clauses or adjust prices in contracts within set limits, but critical decisions—like approving price constraints or validating risks—remain under human control. This hybrid model of AI with human oversight works well in regulated industries like energy, insurance, and telecom.
NTT Data, for instance, worked with Luminance's AI-powered negotiation features to understand the organization's preferred negotiation positions, while customizing how the system defines contracts and clauses. At Maersk, AI grew smarter over time, delivering better price results after more rounds of negotiating with a specific supplier, but the human expert approves the final agreement. In the telecom sector, Vodafone and Deutsche Telekom used semi-autonomous AI systems to negotiate maintenance and operations contracts. Vodafone, which has over 300 million customers, achieved significant savings while maintaining high-quality service, showcasing how semi-autonomous negotiations can deliver real business impact.
Fully Autonomous Stage
These systems handle negotiations end to end—within guardrails. They use real-time inventory, supplier history, and market data to close dozens of deals at once. One example is Walmart's use of AI to negotiate replenishment terms with suppliers for frequently purchased, low-margin items without human approval. Another is Advanced Micro Devices' use of Luminance's Automark-up, an AI-powered tool that can markup legal contracts like non-disclosure agreements autonomously.
How to Make the Journey
Involving human expertise in the transition from AI-assisted to semi- or fully autonomous negotiations not only helps correct potential flaws but also fosters a two-way learning process, enabling both humans and algorithms to continuously refine and enhance their performance. Here are some good practices:
Ensure the quality of the data is high.
When using AI for contract negotiation and drafting, organizations must ensure that data on things like supplier performance, benchmarks, and market trends is accurate, timely, and complies with all local laws. AI negotiates with what it knows. This means that AI should be trained on domain-specific legal data to ensure it produces clear, enforceable contracts aligned with applicable law. Remember: In legal contexts, precision is non-negotiable. Prioritize better data over more data.
Protect data privacy and security.
Using AI for negotiations means sharing sensitive data. To safeguard it, companies need strong protections such as encryption, access controls, anonymization, and regular risk reviews. This isn't just about your own cybersecurity; it's also about building trust with suppliers and regulators.
Establish clear accountability frameworks.
If an AI tool makes a mistake—like incorrectly assessing a supplier—the company, not the software, is responsible. These errors can lead to serious legal or financial consequences. Therefore, companies should establish clear accountability guidelines with defined procedures for review, redress, and oversight. Companies should establish safeguards such as the obligations of parties to disclose the use of AI, how it operates, how the parties' data will be used, and how privacy will be protected. This is especially important in B2B contexts, where regulatory requirements may be less defined.
Stay compliant with regulations.
In many places, regulations such as the European Union's General Data Protection Regulation and the upcoming Artificial Intelligence Act, which was enacted last year and which is now gradually going into effect, require human oversight in sensitive decisions made by autonomous systems. These regulations aim to protect fairness, transparency, and accountability, especially for fundamental rights such as employment, housing, and healthcare. For example, while AI may assist by identifying contracts up for renewal or drafting contracts, it must not send them to suppliers without human approval. An effective way to demonstrate compliance is to track the stages in the process that require human verification.
Build trust through explainability.
It is difficult to trust decisions you don't understand. Use AI models that can explain their reasoning and avoid deploying black-box systems for high-stakes decisions. This transparency is essential for scaling adoption: Both Dell and Walmart reported stronger adoption once their tools showed how decisions were made, not just what they were.
Rethinking the Negotiation Profession
Some worry that automation could hurt career development in areas such as procurement—particularly for junior talent. But we believe such fears are overblown. Reviewing dozens of repetitive contracts doesn't necessarily make someone a better negotiator. Instead, automating those tasks frees up people to spend more time on more strategic, high-stakes negotiations, where human judgment remains essential. The shift to AI will not only drive efficiency, it will also free up talent for more strategic, higher-value tasks.