I agree. The model is only part of the equation now. What really separates these platforms is how well they turn that intelligence into reliable actions, continuous improvement, and a better overall support workflow.

I think this is exactly where AI customer support is headed. The model matters, but once multiple platforms have access to strong LLMs, the real difference comes from everything built around it.

As powerful LLMs become widely available, the model itself is becoming less of a differentiator. The real advantage is likely to come from how well a platform turns that intelligence into actual customer outcomes.

For me what separates them is execution not just the LLM. Better automation, integrations, accuracy, speed, analytics, and the overall customer experience make the real difference.

The LLM is only one part of the product. What really separates platforms us everything around it: knowledge retrieval, integrations, actions, context, guardrails, analytics and human handoff. Two platforms can use similar models and still provide completely different customer experience

The LLM is becoming more of a baseline than the main differentiator. What really matters is everything around it; integrations, accuracy, analytics, action-taking, handoffs, and how easy it is to improve the agent. A slightly less “powerful” model with a better overall system could easily deliver a better customer experience.

the biggest hidden cost of AI support id maintenance . if a platform requires a full time team of prompt engineers just to patch cases

I’d say ease of maintenance is a huge factor too, a platform can look impressive at launch, but if it takes constant manual work to keep the AI accurate and useful, the benefits start to disappear, the best systems should improve without creating another full time job for the support team.