Intuit's decision to scrap its agentic AI architecture twice within four months reveals a fundamental truth about enterprise AI deployment that CX teams need to internalise: architectural iteration at speed is now a competitive necessity, not a sign of failure. By reframing these rebuilds as "the fast path" rather than setbacks, Intuit's VP of AI Nhung Ho signals that the industry has moved beyond waterfall-style AI planning. The company's willingness to abandon specialist agent fleets and restructure its approach suggests that early assumptions about how agentic systems should be organised—assumptions that likely seemed sound during initial design—crumbled under real-world operational pressure. For teams currently evaluating or implementing agentic solutions, this raises a critical question: are your vendor partnerships and internal roadmaps built to accommodate this kind of architectural flexibility, or are you locked into rigid deployment models that will become liabilities as the technology matures?
The implications cut deeper than process. Intuit's experience indicates that success in agentic AI depends less on getting the initial architecture right and more on building organisational muscle for rapid iteration. This contrasts sharply with traditional CX platform implementations, where architectural decisions made during the sales cycle tend to calcify. The Agentic AI Reality Check: Why Trust and Control Are Eclipsing Model Power and Brex built its AI agent policy by watching what agents actually do, not by writing rules first both underscore that control mechanisms and operational guardrails matter more than raw capability—and these can only be discovered through deployment and observation, not pre-implementation planning. For CX leaders, this means budgeting for architectural rework as a line item, not treating it as technical debt. The teams that will win are those prepared to iterate on agent design based on actual customer interaction patterns rather than theoretical models.
What remains unresolved is whether mid-market and smaller CX operations can afford this iterative approach, or whether they'll be forced to adopt battle-tested architectures from larger vendors at the cost of customisation. Intuit had the resources to rebuild twice; most support teams do not. This creates a potential consolidation pressure in the CX stack, where only vendors with sufficient scale can absorb the cost of architectural evolution and pass stable, proven designs downstream to their customers.
Intuit was an early pioneer in the usage of agentic AI, but its path to success has hardly been a straight line.At VB Transform 2026, Intuit VP of AI Nhung Ho described how the company rebuilt its agent architecture twice in the span of about four months, first moving from a fleet of specialist agen