Amazon has chosen Enterprise Connect to launch a set of Amazon Connect updates.
Collectively, the new releases seem to argue that most contact centers are still measuring the wrong things, managing the wrong way, and leaving too much to chance before anything goes live.
The four releases cover:
- Predictive personalization
- AI-assisted management
- Pre-deployment simulation
- Email analytics
The scope is wide, but there is a consistent thread running through all of them. Amazon Connect is pushing further into intelligent operations territory, well beyond its roots as a routing and telephony platform.
Rethinking What AI Success Actually Looks Like
The most ambitious of the four releases is Predictive Insights.
Currently in Preview, the solution is designed to challenge one of the contact center industry's most enduring benchmarks.
As Amazon notes, “most companies measure AI success by how many calls it deflects.”
Predictive Insights is built around the idea that deflection is the wrong goal.
Instead, the feature uses Amazon Connect's unified data layer to pull signals from every touchpoint a customer has had – what they browsed, what they asked in a previous chat, what they called about last week – and converts that data into real-time intelligence that shapes the next interaction before it even starts.
Amazon describes it as turning “customer service from a cost center into a relationship-building engine.”
Amazon is not alone in making that argument at Enterprise Connect this week.
Indeed, Zoom made a similar case with its “resolution economy” framing, challenging the same deflection-led metrics that have dominated contact center measurement for years.
But where Zoom's pitch is largely about redefining how success is measured, Predictive Insights puts forward a specific mechanism: a unified data layer feeding hyper-personalization at the point of contact, giving the argument more operational substance than the typical vendor soundbite.
Supervising AI and Human Agents from the Same Place
Assistant for Managers, also in Preview, addresses a problem that has become more pressing as blended contact centers grow.
AI agent performance and human agent performance are typically tracked in separate systems, which means supervisors are constantly stitching together reports before they can form a coherent view of what is actually happening on the floor.
The feature gives managers a single natural-language interface to query both. Ask which queues are at risk of missing service level targets, and you get an immediate answer with recommended recovery actions, which means no reports, and no toggling between dashboards.