---
id: "concept-agentic-ai-negotiation"
type: "concept"
source_timestamps: ["§ How Companies Are Using Agentic AI to Negotiate"]
tags: ["artificial-intelligence", "automation", "procurement"]
related: ["claim-ai-replaces-routine-negotiation", "entity-walmart", "entity-maersk", "entity-mit", "question-ai-negotiation-ceiling"]
definition: "The use of autonomous generative AI systems to conduct multi-issue contract negotiations within predefined parameters, primarily used to optimize high-volume, routine agreements."
confidence: "medium"
sources: ["ecosystem"]
sourceVaultSlug: "hbr-seg-ecosystem"
originDay: 11
articleStem: "hbr-nm-103-big-companies-negotiate-deals"
sourceUrl: "https://hbr.org/2026/01/why-big-companies-struggle-to-negotiate-great-deals"
sourceTitle: "Why Big Companies Struggle to Negotiate Great Deals"
---
# Agentic AI in Negotiation

Generative AI is presented as evolving from a supporting contract-analysis tool into **autonomous agents** capable of negotiating full contracts with human counterparties or other bots. Per the source, companies like [[entity-walmart-d11|Walmart]] and [[entity-maersk-d11|Maersk]] use AI agents to execute thousands of multi-issue trades within set parameters — particularly for **'tail-end' supplier contracts** that previously wouldn't have been negotiated at all due to resource constraints.

These agents can navigate ostensibly zero-sum issues by matching priorities on payment terms, delivery schedules, and termination clauses. The source cites [[entity-mit-d11|MIT]]'s 2025 AI Negotiation Competition (200+ agents) as evidence that bots can reach multi-issue agreements and that their underlying strategy significantly affects value creation. The strategic payoff: human negotiators focus exclusively on high-stakes deals requiring judgment and creativity (see the [[concept-market-standard-default|market-standard]] logic).

See claim [[claim-ai-replaces-routine-negotiation]] and open question [[question-ai-negotiation-ceiling]].

**Enrichment / confidence — read this carefully.** The *direction* (AI absorbing routine, low-value procurement negotiation) is credible and consistent with research trajectories and CLM vendor roadmaps. But the **specific factual claims are unverified**: as of 2024 there is no independently verifiable public evidence that Walmart or Maersk run *thousands* of fully autonomous end-to-end multi-issue negotiations with external counterparties, nor a public record of an 'MIT 2025 AI Negotiation Competition' with 200+ agents under that name. The real, documented precursor is the **Automated Negotiating Agents Competition (ANAC)**, which shows bots can reach agreements and that strategy shapes outcomes. Treat the corporate examples as forward-looking, anonymized/composite, or speculative. Governance caveats (accountability/liability, explainability/auditability, adversarial-prompt behavior against sophisticated counterparties) argue for retained human oversight.


## Related across articles
- [[concept-ai-layoff-anxiety]]
- [[claim-single-income-risk]]