Meta's infrastructure VP has issued a stark warning: organisations have approximately 20 months to fundamentally rebuild their systems to support agentic AI, or risk obsolescence. This timeline reflects not hyperbole but the accelerating pace at which AI agents are moving from experimental pilots to production workloads. The statement carries particular weight given Meta's infrastructure investments and its position as both a technology provider and early adopter. The underlying message is unambiguous—the shift from traditional request-response systems to autonomous agents operating within enterprise environments represents an architectural inflection point, not an incremental upgrade cycle.
For CX teams, this creates an immediate tension between urgency and pragmatism. Your current stack—whether Zendesk, Freshdesk, or Salesforce Service Cloud—was architected for human-supervised workflows where agents handle tickets sequentially and escalation paths are predetermined. Agentic systems operate differently: they require real-time decision-making authority, access to multiple backend systems simultaneously, and the ability to complete customer requests without human intervention. The question becomes not whether to adopt AI agents, but whether your existing infrastructure can support them without wholesale replacement. Vendors like Oracle are already widening agentic capabilities across their platforms, whilst startups like Cue are raising capital specifically to accelerate agent-based customer service, suggesting the market is bifurcating between platforms that can adapt and those that cannot.
The 20-month window matters because infrastructure transformation at enterprise scale takes time—planning, procurement, integration testing, and staff retraining cannot be compressed indefinitely. Teams that begin architectural assessment now may complete migration by the deadline; those waiting for clarity risk finding themselves locked into legacy systems precisely when competitive advantage shifts to organisations running autonomous agents. The real risk is not that your current platform will disappear, but that it will become a constraint on what your organisation can deliver to customers.
Organizations need to transform to meet the needs of agentic AI.Meta VP of Engineering Barak Yagour opened his talk at VB Transform 2026 wearing a pair of Ray-Ban Meta AI glasses, a small sign of how far AI has already worked its way into physical life. His argument went further: enterprise infrastr