Good week for the people running the floor. Supervisors get a real intervention tool for messaging, and agents get to decide when AI summaries happen instead of waiting for the system. Outbound admins pick up three new Campaign Health signals too. Here's what I think matters.
TL;DR
- Supervisors can now remove an agent from an active messaging conversation and continue with the customer directly. It's gated behind a new permission, so nothing happens until you assign it.
- Agent Copilot can generate an interaction summary on demand, mid-conversation, and the agent can edit the result.
- The Campaign Health view adds recycle events, compliance abandon rate, and calls per agent for voice campaigns.
- Social listening gains automatic explicit-image masking, and Analytics dashboards get a visual refresh with redesigned pagination.
Agent Experience
Supervisor takeover for messaging conversations. Until now, a supervisor watching a messaging interaction go sideways could coach the agent or barge in alongside them. That was it. This release adds a real takeover: a supervisor with the new permission can remove the agent from an active messaging conversation entirely and continue with the customer themselves. The permission is new, so nobody has it until an admin hands it out. Requires CX 1, 2, 2 Digital, 3, 3 Digital, or 4 (Add-on II).
On-demand summaries in Agent Copilot. Agents can now trigger an AI summary of an active interaction with a button press instead of waiting for the automatic end-of-interaction summary. The summary lands in an editable text box, so the agent can fix it while the conversation is still fresh, and it persists after the interaction ends. Two gotchas. Genesys allows one summary per interaction, so generating one on demand means the automatic end-of-call summary is skipped. And on transfers, the summary carries over to the next segment but the agent's edits to it do not. Also check your summary retention window (configurable from 1 hour to 10 days) against your after-call work timeout; if ACW outlasts retention, edits stop being saved.
Outbound
Campaign Health view enhancements. Voice campaign admins get three new data points in the Campaign Health view: recycle events, compliance abandon rate, and calls per agent. These are the numbers you reach for when a campaign is pacing strangely. Is the dialer abandoning above your regulatory threshold? Are agents saturated or starved? Until now you stitched that together from exports. Small change, real value.
Customer Engagement
Explicit-image detection and masking in social listening. Organizations doing social listening can now enable automatic detection of explicit or suggestive images in ingested posts. Flagged images are replaced with a placeholder in the Social Listening Performance and Posts views, users with the right permission can review and unmask them, and flagged posts are not escalated to the agent desktop at all. If your brand monitors public social channels, I'd turn this on. It's an agent-wellbeing and HR-risk control that used to require manual moderation.
Analytics
Dashboard and Performance view refresh. Analytics dashboards and Performance views get updated styling, redesigned pagination controls, refreshed empty states, and updated modals, menus, and widget displays. No data changes, purely cosmetic. But supervisors who live in these views will notice, and any internal training screenshots or documentation you maintain are now out of date.
What this means for your contact center
Think hard about supervisor takeover before you enable it. Yes, sometimes a conversation needs a supervisor to step in, on a regulatory complaint or an escalating customer. But pulling an agent out of a live conversation is a blunt instrument, and your agents will have opinions about how it gets used. Decide who gets the permission, define when takeover is appropriate versus coaching or conferencing, and write that down before anyone uses it in anger. The permission is brand new, so nothing changes until you act. Set policy first.
The on-demand summaries are the quieter win, and probably the bigger time-saver in aggregate. An agent who summarizes during a lull arrives at wrap-up with the work already done. The catch is the one-summary rule: summarize early in a long interaction and you skip the automatic summary that would have covered the whole thing. Brief your agents on when to press the button.
On the outbound side, compliance monitoring gets more immediate. If you operate under abandon-rate regulations (TSR, Ofcom, ACMA), the compliance abandon rate now sits in the same view you use to watch pacing. Check it against whatever external tracking you've built. You might get to retire some of it.
What to check
- Decide your policy for messaging takeover, then assign the new supervisor permission only to the roles that policy names.
- Enable on-demand summaries in Agent Copilot settings if summarization is already on, and brief agents on the one-summary-per-interaction behavior.
- Verify your Copilot summary retention period is at least as long as your after-call work timeout, or edited summaries won't be saved.
- Review the new Campaign Health metrics against your current outbound compliance reporting for any discrepancies.
- If you use social listening, evaluate turning on explicit-image masking and decide who gets unmask permission.
- Update any internal documentation or training material that screenshots Analytics dashboards or Performance views.
Full details in the official Genesys Cloud release notes for June 22, 2026. Questions about how any of this lands in your deployment? Email me.