During Loeb’s AI Summit in New York City on February 11, 2026, the firm's Litigation lawyers spoke with in-house counsel at various companies about the significant but controllable risks associated with replacing legacy enterprise software across finance, HR, manufacturing, supply chain and other business functions with new integrated systems, especially ones that include embedded AI capabilities. The AI Implementation Dispute Resolution roundtable focused on two primary lessons learned from Loeb’s vast experience litigating and resolving disputes concerning failed software implementations.

First, companies must ensure rigorous pre-contract software validation, particularly the AI-enabled capabilities. As enterprise software rapidly evolves, developers are racing to embed AI functionality that is often new to the market. While these capabilities may be compelling in sales demonstrations, the risk of purchasing immature or insufficiently proven functionality is real. This is especially true where AI outputs could influence financial reporting, operational planning and other business decisions. During the sales cycle, companies should require demonstrations using real business processes, validate outputs against the company’s own data and confirm that the exact software modules and versions have been successfully deployed at comparable companies and scale.

Second, companies must ensure that its implementation vendor conducts comprehensive testing before going live with the new system. Experience shows that most implementation failures stem from inadequate testing, including deficient “day in the life” testing, volume and stress testing, and full end-to-end integration testing. It is critical that companies perform parallel runs for business processes such as payroll and financial close, and confirm system performance under peak activity levels. An independent quality assurance review should also be completed prior to go-live to provide an objective assessment of system readiness.

The stakes are too high for companies to implement unproven software or go-live before the system is fully validated. By enforcing rigorous pre-contract validation, robust testing and independent quality assurance, companies can materially reduce the risk of a failed software implementation and the ensuing damages, costs and operational disruption caused by a flawed and defective system.