📊 45% of your team is using AI (secretly?)

Also inside: The bot that prevents empty shelves before customers even notice.

Dec 16, 2025

Start every workday smarter with CX AI news, tools, and data. Become the go-to person on your team for what’s next.

📅 December 16, 2025 | ⏱️ 4-min read

Good Morning!

I’m trying out a new format today. Why? Because your inbox is a war zone. As working practitioners, we don’t need more “content,” we need clarity. I’ve redesigned this issue to be cleaner, punchier, and focused entirely on the “So what?” for your strategy. My goal is to get you the highest signal-to-noise ratio in the industry, in under 4 minutes.

Let me know if it lands. Scroll to the bottom to provide feedback.

Thanks!

☕ The Executive Hook

Most “AI agents” we hear about are still just really good chatbots. But that changed this week. With Zoom and NVIDIA dropping major updates, we’re moving from AI that just talks to AI that actually does the job. If you’ve been waiting for the moment to hand off real, grunt-work tasks to a bot—it’s here.

🤿 The Deep Dive: Zoom just woke up

The News: Zoom launched AI Companion 3.0, and they’re officially jumping into the “Agentic AI” race.

Why I’m paying attention: Until now, your meeting assistant was basically a glorified stenographer. It took notes, maybe summarized a few action items, and then went to sleep.

Zoom’s new update changes the game. These aren’t just passive listeners anymore; they are active executors.

The Shift: Instead of just transcribing “let’s open a Jira ticket,” the agent can now connect to your apps and actually open the ticket without you leaving the video call.
The “Federated” Model: They aren’t betting on just one brain. The system swaps between different LLMs to find the best one for the specific task at hand. Smart.

What this means for you: Think about the friction in your day. You finish a call, then spend 20 minutes updating Salesforce, sending emails, and logging tasks. Zoom is trying to kill that “post-meeting tax.”

For us in CX, imagine a support manager reviewing a call and clicking a single button to “Generate Training Module” based on the transcript. No tab switching. No copy-pasting. Just done.

📊 CX By The Numbers: Who’s really using AI?

The Stat: 45%. That’s how many U.S. employees Gallup says are now using AI at work, up from 40% earlier this year.

The Reality Check: Look around your office (or your Zoom grid). Nearly half the people you see are using AI tools. But here’s the kicker: 10% are using it every single day.

My take: If you think your company is “waiting to roll out AI,” you’re deluding yourself. Your team is already doing it. They’re using it to “consolidate info” (42%) and “generate ideas” (41%).

This is classic “Shadow AI.” If you don’t give them secure, enterprise-grade tools, they’re going to use whatever free tool they found on Twitter. And that’s a data privacy nightmare waiting to happen. Stop policing it and start enabling it.

🛠️ The AI Toolbox: Crisp Retail AI Agent Studio

Moving from “What” to “Why”

The Problem: Retail and CPG teams are drowning in dashboards. They can easily see what happened (e.g., “sales are down 5% in the Northeast”), but they spend countless hours digging through spreadsheets to figure out why.

The Solution: Crisp’s AI Studio allows businesses to build specialized AI agents that autonomously perform Root Cause Analysis.

Automated Investigation: Instead of waiting for a human analyst, the agent detects the anomaly and instantly drills down to find the specific driver (e.g., “Stockout at Distribution Center A”).
Actionable Alerts: It doesn’t just generate a report; it pings the specific team member in Slack or Teams with the diagnosis and a recommended action.

The CX Win: Availability is the foundation of experience. You cannot have a customer experience if the product isn’t on the shelf. This tool shifts CX from “apologizing for out-of-stocks” to “preventing them before the customer even walks in the store.”

Source: Crisp

⚡ Speed Round: Quick Hits

Build your own agents (cheaper): NVIDIA just released the Nemotron 3 family of open models. Why care? It gives your dev team the raw materials to build highly specialized, private agents without paying a massive “token tax” to OpenAI every month. NVIDIA Investor Relations
Bots buying ads: Media agencies are testing AI agents that don’t just plan campaigns but execute the buys autonomously. If we trust AI to spend our marketing budget, how long until we trust it to issue refunds without a manager approval? Digiday
The Google killer? A new forecast says B2B search budgets are shifting to social and CTV because Generative AI search (like ChatGPT) is eating traditional Google volume. If you rely on SEO, your 2026 strategy needs a serious rethink. PPC Land

📡 The Signal

The theme for today is “The Handoff.”

We’re moving past the phase where AI just gives us info (dashboards, summaries). We are entering the phase where AI takes action (Zoom’s execution agents, Crisp’s inventory agents)

My challenge to you: Look at your roadmap for Q1. Do you have a project that involves AI doing something, or just saying something? If 90% of your strategy is “answering FAQs,” you’re solving 2024’s problem. Find one process (refunds, tagging, restocking) and hand the keys to an agent.

See you tomorrow,

👥 Share This Issue

If this issue sharpened your thinking about AI in CX, share it with a colleague in customer service, digital operations, or transformation. Alignment builds advantage.

📬 Feedback & Ideas

What’s the biggest AI friction point inside your CX organization right now? Reply in one sentence — I’ll pull real-world examples into future issues.

Discussion about this post

Ready for more?