Your daily signal on AI and CX — minus the hype.
DCX Stat of the day: About 21% of customer calls follow a failed attempt to solve the issue online, and 94% of those failures happen because the issue was too complex for self-service. ContactBabel
IN THIS ISSUE
→ Why 94% resolved needs a second measurement
→ The call that exposes weak self-service
→ What AI shopping demands from product data
→ A manager copilot for automation decisions
→ Banking tools that move money and risk
🔍 DEEP DIVE
A Success Story With an Operating Lesson
A member asks about a debit card, gets help without waiting for an employee, and completes the task in the channel they already chose. That's the promise behind KeyPoint Credit Union's expanded use of Eltropy: make routine service faster for members while giving employees more room for the work that needs them.
KeyPoint Credit Union says its expanded Eltropy deployment resolved 94% of member inquiries, cut chat volume 67%, and lifted voice service level from 75% to 87% in the first month. The platform now handles voice, chat, appointments, and a new debit-card authentication path.
Taken together, those results point to more than a chatbot win. KeyPoint expanded the same service model across voice, chat, appointments, and debit-card authentication. That's what makes this story useful: the AI is attached to specific member jobs and supported by a broader service design.
Bottom Line: The strongest AI service wins come from connecting the technology to real customer jobs, then expanding where the operating results support it.
📬 Copy-Paste Take
KeyPoint's story is a reminder that AI service gets more valuable when it moves beyond a single channel. Start with a clear member job, connect the surrounding workflow, and expand when the experience and operating results both improve.
🧭 OPERATOR PLAYBOOK
Build From the Journey, Not the Channel
Choose one high-volume customer job where waiting, channel switching, or authentication creates unnecessary effort.
Use four checks to connect the automation to the full journey:
Did the customer complete the intended task?
Did the customer return through another channel?
Did an employee correct, reopen, or explain the answer?
Did authentication or escalation create new effort?
Then compare the AI label with what actually happened across the journey.
Ask your team: Which customer job could improve if voice, chat, authentication, and employee support worked as one system?
Signal: When AI is tied to a specific customer job, teams can make better decisions about where to expand, where to add safeguards, and where human judgment still matters most.
📊 MARKET REALITY CHECK
Product Data Just Became Customer-Facing
Seventy-two percent of 1,028 surveyed U.S. consumers expected retailers to offer AI-assisted shopping within a year. Microsoft also cites Adobe data showing that AI-referred visitors converted 42% better than other traffic in Q1 2026.
That does not prove every retailer needs an agent. It does mean product feeds, prices, availability, returns, and policy language are no longer back-office housekeeping. They shape what an assistant recommends before the customer reaches the site.
Why it matters: Incomplete data can remove a product from consideration, while stale data can win the click and lose the customer at checkout or delivery.
Machine-readable product truth = customer experience
🧰 TOOL WORTH KNOWING
Amazon Connect Customer AI Analytics
What it does: Amazon Connect Customer AI Analytics lets managers ask questions in plain language across more than 150 self-service, agent, and queue metrics. It returns an answer, supporting evidence, and a recommended action.
CX use case: A service leader can ask which queues are best suited to automation and see a prioritized list based on handle time and after-contact work.
Worth watching because: The tool moves AI upstream, from answering customers to advising managers where automation should go next. That makes metric definitions, missing data, confidence scores, and managerial review part of journey design.
Bottom line: Treat the recommendation as a hypothesis. Check repeat contact, vulnerable-customer signals, exceptions, and recovery before turning it into a roadmap decision.
The DCX AI Today - AI Tool Directory - If you lead a CX team and want a curated shortlist of tools worth evaluating, this is your starting point.
📡 90-SECOND CX RADAR
Starling Gives Its Banking Assistant Bounded Jobs
Starling is rolling out Smart Tools for specific money tasks, beginning with a Rainy Day Saver and Fraud Controls. The savings tool can analyze income and direct debits, create a savings space, and automate transfers.
Why it matters: Bounded actions make an assistant easier to understand, but weekly releases create a fast governance cycle around consent, affordability, transfer limits, fraud settings, explanation, and reversal.
✅ YOUR MOVE
Pick one AI metric your team celebrates: resolution, containment, conversion, deflection, or time saved. Then find the customer behavior that could contradict it.
Spend 30 minutes tracing 10 cases past the point where the dashboard stops. Name the missing measure, assign an owner, and review it again in two weeks.
The customer journey keeps going after the metric declares victory.
Until tomorrow,
👥 Share This Issue
Think of one person who’s wrestling with AI in CX right now
and forward this to them.
I’m obsessed with Wispr Flow Pro! Get a Free Month on me.
If someone forwarded this to you, they thought you needed to see it before your next AI planning meeting. Get your own copy.