What changed
AWS announced general availability of Resolutions in Amazon Connect Customer, an agentic solution for high-volume operational work such as refund requests, transaction disputes, fraud investigations and insurance claims. Managers upload a standard operating procedure or describe the process in plain language; Resolutions asks clarifying questions about exceptions, decision rules and approval thresholds, then generates a Work plan of Phases and Steps. Publishing a Work plan generates the flows, Case template, Case fields and Task templates it needs. Each incoming request, including requests started from inbound email, becomes a Resolutions Case. Steps can run on prebuilt AI agents, AI agents configured in Connect or your own agents, call business applications through flow modules, email customers for missing details, and route approvals to specialist queues. Every action and decision is recorded on the case, and operations managers get status and analytics per Work plan.
Vendor documentation (opens in a new tab)Why it matters
This pushes Amazon Connect beyond conversations into back-office case resolution, the work that usually stalls between teams and applications after the contact ends. The value depends on how well your SOPs encode exceptions and approval limits, and on which actions you let AI agents take in connected systems, so governance and audit design matter as much as the build.
CxCave practical context
Who should care
- Architects
- CX leadership
- Administrators
- Developers
Implementation considerations
PLANNING RECOMMENDEDPick one well-documented, high-volume process such as refunds, generate a Work plan in a test instance, and review the generated flows, Case fields and approval steps with the process owner. Decide which business-application actions AI agents may take without approval, and check Resolutions pricing before scaling.
CxCave recommendation, not a vendor mandate unless explicitly stated.
