front-end developments

🖤 we're loving: Understated AI with overachiever energy

🔮 we'll see: Is AI getting a little too humanlike? Brb, paging Freud

⚠️ say less: Duolingo’s AI tried replacing people—turns out, people noticed

back-end breakdown

skip the theatrics of performative AI—quiet, competent AI already delivered

This month offered a masterclass in what separates effective AI from performative adoption. The takeaway? When AI is invisible, intuitive, and embedded into real outcomes, it builds loyalty. When it’s loud, rushed, or vaguely justified as “the future,” customers tune out or push back. 

🔢 semantics:

• Bank of America’s Erica hit 3B interactions, earning praise for its fast, accurate, and user-first approach to AI in banking

• Zoom’s contact center growth (94% YoY) reflects strong uptake of AI designed to support—not replace—human agents

• Duolingo’s growth slowed after positioning AI as a contractor replacement, sparking backlash and eroding user trust

📈 sentiment analysis:

You don’t earn trust with “disruption.” You earn it with design. Customers don’t care that you use AI—they care how it shows up. Is it helpful? Does it actually improve the experience? Being AI-first doesn’t mean replacing people. The best AI builds trust by working with them. Quiet competence wins where loud strategy doesn’t.

ada recommends

your next star service rep won't ask for a desk

AI support coverage across channels

rethinking AI strategy in customer experience

old metrics die hard. yippee-ki-yay, NPS

CX leaders are questioning whether legacy metrics like NPS and CSAT are still pulling their weight. Static scores can’t keep pace with real-time, dynamic experiences and often miss the moments that matter most. The tools we’ve relied on for decades might not be broken, but they’re definitely outdated.

🔢 semantics:

• NPS and CSAT are increasingly criticized for measuring intent and sentiment—not actual customer experience quality

• Predictive metrics like xNPS estimate satisfaction mid-interaction, enabling live recovery actions

• AI dashboards are replacing static reporting with real-time, actionable insights across the full customer journey

📈 sentiment analysis:

We’re not saying NPS belongs in a museum, but it shouldn’t be driving your roadmap. When AI is resolving issues faster than customers can hit “submit” on a survey, you need customer service metrics that move at the same speed. 

🤝 “Emotionally intelligent AI builds trust.”

Some think the future of customer experience lies in AI that responds with emotional awareness. These systems don’t just process queries, they read the room. By adapting to user tone, context, and frustration in real time, emotionally perceptive AI earns trust, de-escalates tension, and boosts loyalty. It’s not about mimicking people, it’s about understanding them.

⚠️ “Pretending to be human crosses a line.”

Others—like Mustafa Suleyman—argue that when AI starts to seem sentient, things get murky fast. Users project emotions, assign rights, and form attachments to systems that aren’t actually alive. The line between good UX and ethical confusion? It’s thinner than you think—and we’re not ready for what’s on the other side.

⚖️ our take:

We’re all for AI that’s emotionally aware. But let’s not confuse empathy with identity. The best AI doesn’t pretend to be human, it understands how to support one. Keep it intelligent, keep it intuitive, but let’s also keep it grounded. Clarity builds trust—confusion breaks it.

🧠 GPT-5 debuts

OpenAI’s latest model is smarter across 40+ job domains. The AGI arms race is back on.

🤖 Aisle be back

Walmart’s AI is now running the floor, executed with chilling accuracy. Hasta la vista, inefficiency.

🚨 Risky business

25% of execs admit to pasting sensitive info into AI tools. Governance? Never met her.

📱Apple phone home

Apple’s in talks to power Siri with Google’s Gemini. The update? TBD. The existential crisis? Already downloading.