Developing
Most numbers come from vendors (NiCE, Five9, Pindrop) with commercial interests and a short observation window.
In 60 seconds
- NiCE says Meta's Muse reached No. 1 among free US App Store apps a week after launch and 2.8 million downloads in 12 days.
- Pindrop says weekly Muse call volume is 8x its first-week level; 47% of Muse calls use keypad IVR navigation and 64% reach a human agent.
- Pindrop counted 727,000 confirmed AI agent interactions over two months, 94% labelled "good" and 6% "bad".
- Gartner expects 50% of firms that cut service staff for AI to rehire by 2027.
Impact Layer
- What changed
- Contact centres now see consumer AI agents calling them on customers' behalf, at scale.
- Why it matters
- If agents call for customers, call volumes, IVR design and fraud controls all change, and blocking or welcoming them is a policy choice.
- Who is affected
- Contact-centre operators, fraud teams and CX leaders.
- Martech stack area
- AI, Customer Experience
- Stack impact
- Decide your policy for AI agents that call or chat, and check whether your IVR and authentication handle them.
- Business impact
- Moderate Partly opinion. Figures come from NiCE, Pindrop, Five9 and a Gartner survey of 321 respondents; Pindrop's window is short and how "good" and "bad" were labelled is not explained.
- Related technologies
- What to watch
- Independent measurement of agent-initiated contacts and rehiring trends.
CX Today’s Rob Wilkinson argues that Meta’s Muse agent may grow, not shrink, contact-centre work. The page shows no explicit date; related items are dated October 14-15, 2026.
The figures (vendor-reported)
- NiCE: Muse reached No. 1 among free US App Store apps one week after launch and 2.8 million downloads across the US and Canada within 12 days.
- Pindrop: weekly Muse call volume is 8x its first-week level, about a month later.
- Pindrop: 47% of Muse calls use keypad IVR navigation and 64% reach a human agent; median time is 2:01 in the IVR menu and 2:18 with a human.
- Pindrop: 727,000 confirmed AI agent interactions over two months, 94% labelled “good” and 6% “bad”.
- Gartner: 50% of firms that cut service staff for AI are expected to rehire by 2027; only 20% of leaders reduced staffing for AI.
Quotes
NiCE’s Neeraj Verma: “Neither blocking nor welcoming is a strategy on its own.” Five9’s Jay Lee: “If an AI agent will sit on hold for you, a lot of calls that never got made start getting made.”
Caveats
Most figures come from vendors with commercial interests, Pindrop’s window is short, and how “good” and “bad” bots were labelled is not explained. Gartner’s staffing numbers come from a 321-respondent survey, and parts of the article are opinion.
Our read
Agent-initiated contact is real and measurable now. Set a policy before volume forces one.
Sources
- CX Today: Muse Shows Why AI May Grow the Contact Center Again (opens in a new tab)
Source for the vendor figures, quotes and caveats.