Legal AI alliances – is it time to negotiate better terms?
Jinfo Blog

By Stephen Phillips

Abstract

Item

The legal information market is undergoing a seismic shift.

The convergence of practice-management tools, global legal intelligence, and agentic AI, is fundamentally reshaping how legal research and workflows are delivered, presenting information managers with both risks, and opportunities.

2025 has seen several vendors start to embed AI into legal research platforms:

• Clio's acquisition of vLex
• LexisNexis's alliance with Harvey
• Thomson Reuters' launch of Westlaw CoCounsel.

Will these strategic moves reinforce the long-standing LexisNexis/Westlaw duopoly, or will they create credible challengers?

Meanwhile, information managers from larger law firms continue to be frustrated with vendors, as they continue to refuse their clients' requests to licence content for use in their clients' own AI stack.

Whilst this is frustrating, these developments create leverage and may help encourage vendors to offer more flexible licensing options.

The new in-depth Jinfo Report "Legal research – AI alliances, opportunity or risk?" includes the full spectrum of opportunities, risks, and negotiation levers available to information managers right now.

Here are two of them:

1. The AI workload paradox

Vendors are committed to delivering AI-driven tools, often augmented with agentic frameworks. Yet, despite the hype surrounding efficiency gains, information managers must confront the "workload paradox".

While AI outputs may appear fluent, they mask inaccuracies, distort precedent, or invent citations altogether.

Because legal contexts demand precision and traceability, any legal research performed with AI must be carefully checked. This enhanced checking effectively eliminates the efficiency gains initially promised by AI.

Many Jinfo Subscribers report significant upticks in workload, in spite of the introduction of AI. So, the efficiency gains suggested by vendors to justify product price increases should be carefully examined and are unlikely to be realised.

Ultimately, AI should be regarded as a cost of doing business for the vendors, and not a premium add-on; especially when it requires additional oversight and validation.

2. Link pricing to performance and risk

The inherent risk of inaccuracy of AI provides negotiating levers for information managers, who can (and should) use these conditions to demand performance accountability and better pricing.

Link pricing directly to the performance of the AI technology by negotiating around metrics such as:

• Hallucination Rates: given the significant reputational risks associated with AI hallucinations, require vendors to provide citation accuracy metrics.
  
  
• Citation Fidelity: require vendors to correctly identify real cases, statutes, and regulations, and ensure they preserve the procedural posture and holding of the cited case.

You can negotiate discounts or service credits tied to documented hallucinations and agreed citation error thresholds.

Conclusion

AI will not break the duopoly overnight. If anything, the incumbents are more likely to try to reinforce their dominance by embedding these capabilities directly.

Making vendors accountable for the performance risk of their own AI deployments, whilst pushing back on the commercial implications, may cause vendors to rethink how best to leverage their investments in the technology.

To understand the full spectrum of opportunities, risks, and negotiation levers available to legal information managers right now, read the detailed Jinfo Report, "Legal research – AI alliances, opportunity or risk?".

As always, we welcome your feedback on our resources, via the Jinfo contact page.

« Blog