As AI moves deeper into customer-facing operations, CX leaders face a harder question than whether the technology works: what happens when the model behind it suddenly changes, disappears, or becomes too expensive to use?
In this CX Today interview, Nicole Willing speaks with Ashish Nagar, Founder and CEO of Level AI, about the enterprise risk created by dependency on public AI large language model (LLM) providers. The risk was brought into focus when Anthropic placed restrictions on its new frontier models after the U.S. government imposed export control requirements, prompting concern about continuity, compliance, and operational resilience.
For Ashish, the lesson is direct: enterprises using AI in critical workflows need more control over the systems they rely on.
“If you’re a CIO or business leader looking at AI for critical workflows, you’ve got to look for vendors that have their own models and are not fully dependent on third-party providers. Ask whether they have vertical-specific models, whether their costs depend on OpenAI or Anthropic, what their security protocols are, and whether the model gets better with every use.”
That matters because AI is no longer sitting at the edge of the business. In customer experience, it is increasingly being used in voice agents, contact center analytics, automation, and real-time support. If an underlying model becomes unavailable, the impact can be immediate: paused workflows, contact volume spikes, slower responses, brand damage, customer churn, and pressure on employees who have to absorb the disruption.
