The July 20, 2026 release is all new features, no removals or deadlines this week. Three items are worth your time: a long-overdue AMD configuration option and two additions for teams running Agent Copilot. Here's my read.
TL;DR
- Answering machine detection is now configurable on outbound campaigns. You control when call analysis timers start and how long the system waits before acting on a detection or silence threshold.
- Agent Copilot gets Intent Health Insights: intent health scores and utterance-level fixes, prioritized by impact, to clean up your NLU training data.
- A new Copilot Usage Analytics dashboard gives admins query volume, active users, and usage trends, broken down by AI agent and tool.
Customer Engagement
Configurable answering machine detection
Outbound AMD has been a black box for a long time. The system decided when to start analyzing a call and how long to wait before declaring a live person, a machine, or silence, and you lived with its judgment. If that judgment was wrong for your dialing patterns or your region's rules, tough.
That opens up now. Administrators can configure AMD on outbound campaigns: when the call analysis timers begin, and how long the system waits before it acts once a detection or silence threshold is hit. Genesys frames this as helping you meet regional regulatory requirements and cut down on unwanted silent or delayed calls.
This is a good change. Silent and abandoned calls are a compliance liability in a lot of jurisdictions, and handing admins direct control over the timing is the right fix.
Employee Productivity
Intent Health Insights for Agent Copilot
If you run Agent Copilot, the quality of its suggestions rides on the quality of your intent training data, and that data rots quietly. Utterances drift, intents start overlapping, and nobody notices until the bot's routing gets sloppy.
Intent Health Insights goes after that directly. It surfaces intent health scores plus utterance-level recommendations, and it flags the specific problems that degrade an NLU model: intent conflicts, utterances that are too short or too long, and outliers that don't belong. The recommendations come prioritized by impact, so you work the intents and utterances most likely to move performance first instead of guessing where to start.
The value is clear if you've ever stared at a sprawling intent library with no idea which entries were dragging performance down.
Copilot Usage Analytics dashboard
The second Copilot addition is an admin-facing analytics dashboard. It reports total queries, active users, and usage trends over time, with breakdowns by individual Copilot AI agent and by tool.
This is the adoption-tracking piece that's been missing. Rolling out Copilot without usage data means you can't tell whether agents lean on it or quietly ignore it, and you can't defend the spend at renewal. Query volume and active-user counts by agent and tool give you that picture.
What this means for your contact center
The AMD change is the one with teeth. If you run outbound voice and you operate anywhere with rules on silent or abandoned calls, this moves a compliance lever from Genesys's discretion into your hands. Get your dialer or compliance owner into the campaign settings, see what the timers allow, and tune them against how your specific campaigns pace. Don't treat the defaults as automatically compliant for your region.
The two Copilot items only matter if you're running Agent Copilot, but if you are, use them together. Intent Health Insights tells you where your NLU training data is weak; the usage dashboard tells you whether agents are using the assistant those intents feed. Cleaning up training data nobody consults is wasted effort, and pushing adoption of an assistant built on bad intents just spreads the problem. Read the two dashboards side by side.
None of this carries a deadline, so there's no fire drill. These are opt-in improvements you adopt when you're ready.
What to check
- Open your outbound campaign settings and review the new AMD timer and threshold options.
- Have your compliance owner validate the AMD timing against your region's silent- and abandoned-call rules rather than trusting the defaults.
- Run Intent Health Insights against your Agent Copilot intent library and work the highest-impact recommendations first.
- Open the Copilot Usage Analytics dashboard, check active users and query volume by agent and tool, and use it to judge whether adoption justifies the license.
Full details are in the official Genesys Cloud release notes for July 20, 2026. Questions about how any of this lands in your environment? Email me.