AI Is Your Sourcing Team’s New Negotiation Partner

Jalen Thibou

Geraint John

July 01, 2026 • 6 min read

---

Back in 2018, the CPO of a multinational food company said something that now feels less like a provocative prediction than a prophecy: “The whole process of getting to a fair price, finding the right supplier, is going to end up being an algorithm, not a negotiation.”

Eight years later, the theoretical has become operational, thanks to generative and agentic AI. In our most recent sourcing survey, more than half of CPOs (55%) told us AI is already adding “significant value” in autonomous sourcing, while just over one-third (35%) said the same about negotiations. And these numbers jump by around 20 percentage points when CPOs are asked to project that value forward to 2028.

Copilots Rather Than Captains 

These findings underline the fact that CPOs expect AI to play a progressively greater role during this period in two of the most important sourcing “jobs to be done”: finding the best suppliers, and ensuring a competitive cost base.  

The majority are happy for AI agents to take over tactical and transactional sourcing work from their team, according to our survey data and to Zero100 members we’ve discussed this issue with. But when it comes to negotiating deals with suppliers, they are typically more circumspect: seeing agents as copilots – supporting human buyers with market intelligence and points of leverage – rather than captains in full control. 

The notion of “autonomous negotiation” – as conceived by our CPO in 2018 – is not yet widely accepted, let alone adopted, and there is a lack of clarity about what it means, where it works, and what it still can’t do. 

What the Technology Can Do Now 

Autonomous negotiation exists on a spectrum, from structured bidding automation to conversational supplier-facing agents. The vendor landscape reflects this spectrum. For example: 

• Pactum leads in conversational negotiation at scale, using chat-based AI agents to negotiate directly with suppliers across variables such as price, payment terms, rebates, and delivery within buyer-defined guardrails. 
• Arkestro stands out for predictive anchoring, using AI, game theory, and supplier behavioral data to set a target price before the first quote is submitted, shaping the negotiation from the outset rather than mimicking a human back-and-forth.  
• Keelvar excels at structured sourcing and award optimization, particularly in freight, logistics, and other categories where evaluating large volumes of bid scenarios creates advantage.  
• Fairmarkit focuses on tail spend and routine RFQ automation, combining supplier matching, event creation, and threshold-based award execution to move high-volume sourcing work with minimal manual intervention. 
• Procure AI has a deliberately structured human-in-the-loop architecture, combining autonomous, collaborative, and ambient agents to support tactical negotiations and governed deployment.   

Where People Remain Essential  

AI’s negotiation capabilities will undoubtedly develop, but today’s tech solutions cannot replicate the human qualities that are required during more complex sourcing events. These include:  

• Judgment: People still need to own the calls where the “right” answer is contextual. Agents can negotiate within preset guardrails, but they don’t reliably make strategic judgment calls where the answer depends on unstated context, competing priorities, or ambiguity. 
• Relationships: People still need to manage supplier trust, history, and long-term value. Agents can run structured exchanges, push for better terms, and process high-volume supplier interactions, but they do not truly understand relationship context. 
• Escalations: People still need to step in when the negotiation leaves the expected path. Among current tech solution providers, escalation is a core design feature that upholds human-in-the-loop for negotiation edge cases. 

What This Means for Sourcing Leaders 

While agentic systems are unable to fully replicate human judgment, supplier relationship awareness, or calculated tradeoffs, they can run high-volume, repetitive negotiations within clear guardrails – often better and faster than a human team could at scale.  

In recent weeks, Zero100 community members whose digital sourcing strategies we’ve reviewed have all included tail spend automation as a priority use case under the banner of “autonomous” initiatives. It’s seen as a relatively low-risk starting point for what agents can deliver, with high volumes, bounded complexity, and acceptable tradeoffs defined in advance.   

This also reflects the fact that, as demonstrated at pioneer Walmart, the long tail of indirect goods and services spend has often not been professionally sourced before – hence it’s loaded with the sort of cost-saving opportunities CPOs need to demonstrate AI ROI. 

Three practices to help capture these opportunities are:  

1. Redesign workflows before automating them. Autonomous negotiation won’t create value if it’s layered onto a sourcing model that still assumes humans do every handoff, exception, and escalation. The first move is to decide which negotiation activities should be fully machine-run, which should be machine-supported, and which should remain human-led – then redesign roles around that split.  

2. Set guardrails, thresholds, and human controls before picking a vendor. When you deploy an agent, you are putting your company’s reputation in its “hands.” Leaders should explicitly define which categories, spend thresholds, and decision types can be delegated, which require approval, and what must trigger escalation before they enter a tech selection process. Although software vendors handle governance differently across the spectrum, they design it as a prerequisite. 

3. Fix the decision-grade data layer in parallel with the pilot. If supplier master data is fragmented, spend definitions vary across systems, or contract and terms data cannot be trusted, you run the risk of AI making bad decisions faster. Identify the critical data that agentic workflows require for negotiations, and clean, define, monitor, and assign ownership for that data first.  

Algorithms have yet to supplant professional buyers from the negotiating role they love – and may never fully do so. But AI is quickly establishing itself as an essential team member, whether acting autonomously or augmenting your people to drive faster, better outcomes. 

Time to make negotiation agents part of your hybrid workforce for the future.